287 62 5MB
English Pages 362 Year 2009
Structural Renovation in Concrete
Structural Renovation in Concrete
Zongjin Li, Christopher Leung and Yunping Xi
First published 2009 by Taylor & Francis 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Taylor & Francis 270 Madison Avenue, New York, NY 10016 Taylor & Francis is an imprint of the Taylor & Francis Group, an informa business This edition published in the Taylor & Francis e-Library, 2009. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2009 Zongjin Li, Christopher Leung and Yunping Xi All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. This publication presents material of a broad scope and applicability. Despite stringent efforts by all concerned in the publishing process, some typographical or editorial errors may occur, and readers are encouraged to bring these to our attention where they represent errors of substance. The publisher and author disclaim any liability, in whole or in part, arising from information contained in this publication. The reader is urged to consult with an appropriate licensed professional prior to taking any action or making any interpretation that is within the realm of a licensed professional practice. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Li, Zongjin, Dr. Structural renovation in concrete / Zongjin Li. p. cm. Includes bibliographical references and index. 1. Concrete construction—Maintenance and repair. 2. Buildings, Reinforced concrete—Maintenance and repair. I. Title. TA683.L48 2009 B624.1′8340288—dc22 2008032399 ISBN 0-203-93136-X Master e-book ISBN
ISBN 10: 0–415–42371–6 (hbk) ISBN 10: 0–203–93136–X (ebk) ISBN 13: 978–0–415–42371–7 (hbk) ISBN 13: 978–0–203–93136–3 (ebk)
Contents
1
List of figures List of tables Preface
vii xi xii
Introduction
1
1.1 1.2 1.3 1.4 2
Degradation of reinforced concrete structures 2.1 2.2 2.3 2.4
3
Building and infrastructure degradation 2 Common causes of structural degradation 2 The objectives and scope of renovation engineering 5 Useful definitions 6
Degradation caused by non-uniform dimensional changes 9 Degradation caused by repeated loading 19 Degradation caused by lack of durability 25 Degradation caused by disasters 62
Inspection and evaluation 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9
8
Preliminary investigation 79 Detailed investigation and evaluation 83 Non-destructive tests 86 Reflected and transmitted waves 118 Concrete strength assessment 121 Surface cracking measurement 127 Assessment of fire damage to concrete 129 Bridge assessment 132 Reinforced steel corrosion assessment 138
76
vi
Contents
4
Conventional repair and strengthening techniques
147
4.1 Principal considerations of repair and strengthening techniques 148 4.2 Repair materials 152 4.3 Repair techniques 177 4.4 Strengthening techniques 235 5
Glass fiber reinforced plastics components for bridge deck replacement 5.1 5.2 5.3 5.4 5.5 5.6
6
243
Introduction 243 Materials 244 Fabrication process and example systems 245 Analysis of FRP bridge deck members 251 Analysis of composite sections 273 Summary 277
Strengthening of reinforced concrete structures with fiber reinforced polymers
278
6.1 Introduction 278 6.2 Structural strengthening with bonded fiber reinforced polymer (FRP) 280 6.3 Flexural strengthening of beams 290 6.4 Design of beams strengthened in flexure 301 6.5 Shearing strengthening of beams 314 6.6 Strengthening of concrete columns 321 6.7 Summary 327 Bibliography Index
328 342
Figures
2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10
2.11 2.12 2.13 2.14 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11
A bridge deck heated by the sun Phenomenon of water bleeding Formation of plastic shrinkage crack Normalized fatigue strength as a function of number of loading cycles Different types of cyclic loadings Hysteresis loop during loading and unloading Chloride diffusion test Diffusion test results Reinforcing steel corrosion and expansion of corrosion products Relationship between the size of capillary pores and the temperatures at which ice formation is possible inside their pores Deterioration of a concrete structure in sea water Earthquake location notation Base isolation based on friction Base isolator Typical layout of ultrasonic testing equipment Sensor configurations in ultrasonic testing Embedded piezoelectric composites as ultrasonic transducers Development of wave velocity in cement-based material Principle of the impact–echo method The main elements of a typical AE instrumentation system Reinforcement potential measurement by half-cell method A non-contact resistivity test apparatus for cementbased materials Four-probe resistivity test of concrete Basic circuit for electrical resistance probe technique Principle of operation of induction meter used to locate reinforcement
10 11 13 20 21 22 33 34 44
57 60 66 74 75 91 92 94 94 96 99 103 104 105 107 108
viii List of figures 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10
4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18
Surface temperature distribution of a debonded tile sample One-dimensional wave propagation at an interface between two mediums The reflection and transmission coefficients vs. acoustic impedance A core-cutting drill A cutaway schematic view of the Schmidt rebound hammer Surface wave velocity measurement for concrete strength assessment Some typical crack types Reinforcement corrosion detection by using linear polarization resistance Corrosion test of reinforced concrete Accumulated AE events as a function of time superimposed with a measurement of galvanic current The coefficient of thermal expansion of concrete vs. temperature Stress distribution in two materials with different moduli of elasticity Bond strength test methods Compressive-shear loading configuration in ASTM C882 Determination of tension softening diagram from the bending test Pull-out tester made by Proceq Example of steel corrosion Application of shotcrete for repairing a concrete arch Many small patches applied on a bridge deck (a) A good surface patching on a concrete slab; (b) a good quality of bonding between patched and old concrete Appearance of a repaired concrete by improper crack repair techniques A concrete core is taken from a bridge deck Ultrasonic testing of concrete cracking before repair Repair of a crack by stitching of concrete Relative ultimate residual strength vs. maximum temperature Relative initial tangent modulus vs. maximum temperature Stress–strain curves of concrete exposed to high temperatures Concrete specimens exposed to maximum temperature 800 °C and cooled down by natural cooling
114 119 120 122 125 128 139 143 145 145 158 159 166 167 168 169 178 189 192
195 198 202 203 204 205 206 206 207
List of figures 4.19 4.20 4.21 4.22 4.23 4.24 4.25
4.26
4.27 4.28 4.29 4.30 4.31 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5.15
Compressive strength of siliceous aggregate concrete at high temperature and after cooling Compressive strength of carbonate aggregate concrete at high temperature and after cooling Strength of flexural reinforcement steel bar and strand at high temperatures Removal of the concrete under corroded steel bars using hand tools Steel bars after cleaning operation Damages and scratches on the epoxy coating of new steel bars A sacrificial anode cathodic protection system, called Galvashield cathodic protection system was installed in SH 85 SB in Greeley, Colorado (a) Impressed current cathodic protection system installed in a concrete slab; (b) schematic for reinforced concrete cathodic protection The control box for an impressed current cathodic protection system Electrochemical realkalization Electrochemical chloride extraction The electrochemical repair system used in the highway industry (a) Severe “D” cracking in a concrete pavement; (b) map cracking in a concrete pavement The pultrusion process for composite fabrication Examples of pultruded GFRP components The filament winding process The hand lay-up process The Manitoba deck system Illustration of the VARTM The Hardcore Composite system The composition of an FRP structural member Local coordinates for the lamina Definition of local and global coordinates Illustration of the plane-section-remains-plane assumption Definition of forces and bending/torsional moments for the plate Stress distribution in the upper half of the [02/±30]s laminate Interlaminar failure between opposite angle plies Physical explanation for the presence of interlaminar shear stresses
ix 208 209 209 213 213 214
216
217 219 220 221 224 231 246 247 248 249 249 250 250 251 253 254 256 257 264 272 273
x List of figures 5.16 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15 6.16 6.17
6.18 6.19 6.20
6.21 6.22 6.23 6.24
Definition of dimensions for the I-section and boxsection Flexural strengthening of concrete beam with bonded FRP Shear strengthening of concrete beam Strengthening of concrete column Schematics of an automatic FRP wrapping system Locations with stress concentrations along the FRP/ concrete interface Debonding failure at the plate end Crack-induced FRP debonding Definition of geometrical parameters for the strengthened beam Transition of failure mode with increasing cut-off distance Effect of tpEp/d and a/S on the failure mode Graph for the determination of (εpu)ref in terms of Eptp/d Comparison of normalized ultimate moment obtained from empirical approach and experimental results Initial condition of the beam Determination of balanced plate ratio for simultaneous steel yielding and concrete crushing Determination of balanced plate ratio for simultaneous FRP rupture/debonding and concrete crushing Determination of ρf,cy under two different situations (a) Debonding failure for a beam strengthened by U-jacketing; (b) FRP rupture for a beam strengthened by full wrapping Effect of multiple cracks on the straining and debonding of FRP strip around the middle of the shear span Definition of terms in the design equations The stress–strain behavior of unconfined concrete and concrete under constant and continuously increasing confinement Failure of an FRP-confined circular concrete specimen by FRP rupture Relation between FRP stress and confining pressure Confinement of concrete column by discrete FRP strips A bilinear model to describe the complete stress–strain relation of FRP-confined circular concrete columns
275 282 283 285 287 291 292 293 294 295 297 300 301 304 306 307 308
315 316 320
322 323 324 325 327
Tables
2.1 2.2 2.3 2.4 3.1 3.2 3.3 3.4 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5.1 5.2 6.1 6.2 6.3 6.4 6.5 6.6
Minimum periods of curing and protection Reduction of permeability of cement paste Limit on chloride ions Classification of severity of sulfate environment according to ACI 201.2R-92 Proposed inspection intervals by FIP General guides to interpretation of electrical test results Survey on behavior of commonly used building materials at high temperature Interpretation of linear polarization resistance measurement results The maximum allowable shrinkage values for repair mortar Factors affecting drying shrinkage Effects of aggregate type on thermal expansion coefficient Moduli of elasticity of repair materials Performance requirements of repair materials in ASTM C928 European Standard EN1504: the new approach to concrete protection and repair Standards of surface preparation Exposure conditions and tolerable crack width Hot rolled reinforcing steel Properties of the fiber and matrix materials for the fabrication of GFRP Summation over all the plies Typical properties of fiber reinforced polymers and steel Effect of cut-off distance on strengthening effectiveness Effect of plate stiffness on strengthening effectiveness Effect of member size on strengthening effectiveness Effect of plate width on strengthening effectiveness Partial safety factor for the FRP
14 29 42 53 78 103 130 143 155 155 157 160 164 165 182 197 210 244 261 281 294 295 297 298 303
Preface
In the past few decades, the deterioration of buildings and infrastructures has been occurring at an ever increasing rate. This has resulted in a worldwide need for the maintenance, repair and rehabilitation of degraded structures. To meet the needs, available knowledge and technologies must be collected, organized and disseminated to engineers and practitioners. Poor understanding of the deterioration problem and the inability to adopt proper remediation approaches not only lead to a waste of natural resources, but also have a negative impact on the economy of the whole world. The maintenance, repair and rehabilitation of concrete structures involve several broad issues which encompass technical, social, and economic aspects, such as the fundamental knowledge of materials and structures as well as basic understanding of economy and sociology. In this book, we focus on the technical aspects. From material and structural points of view, a project for maintenance, repair and rehabilitation of a concrete structure usually requires knowledge and expertise in the areas of structural materials, repair materials, structural inspection, material testing, repair and/or strengthening techniques. To the best of the authors’ knowledge, a book that systematically covers all these issues is currently not available. The present book can be considered the first attempt to fill such a void. The book is divided into six chapters. Chapter 1 gives a brief introduction to renovation engineering. Chapter 2 explores the causes of deterioration of concrete structures, covering durability issues and disaster factors. Chapter 3 discusses the techniques of inspection, including destructive and nondestructive methods. Chapter 4 focuses on the traditional repair and strengthening methods such as crack repair. Chapter 5 covers the application of glass fiber reinforced plastics components for bridge deck replacement, and presents the fundamental mechanics essential for component analysis and design. Chapter 6 provides updated knowledge on the strengthening of concrete structures with fiber reinforced polymers. The book is designed and written primarily to meet the teaching needs for undergraduate students at senior level and graduate students at entry level. It can also serve as a reference or a guide for professional engineers in their practice.
Preface
xiii
In the process of writing this book, the authors received enthusiastic help and invaluable assistance from many people, which was deeply appreciated. The authors would like to express their special thanks to Dr. Xiangming Zhou, Dr Garrison C. K. Chau, and Mr. Xiangyu Li. In particular, Dr. Xiangming Zhou made a sterling effort in collecting information and drafting some parts of the book. The support from China Basic Research Grant, Basic Research on Environmentally Friendly Contemporary Concrete (2009CB623200) is greatly acknowledged. Finally, we would like to thank our wives, Xiuming Cui, Garlin Lee and Xiaoxin Cui. This work is dedicated to their understanding and support. Zongjin Li, Christopher Leung (Hong Kong, China), Yunping Xi (Boulder, CO, USA)
1
Introduction
Renovation is a carefully designed series of activities to recover the load-carrying capability, enhance performance, and extend the service life of existing buildings and infrastructures with a satisfactory quality. Such activities include the repair, strengthening, and rehabilitation of aged or damaged structures. Renovation engineering is a combination of maintenance, inspection, and rehabilitation including repair and strengthening. Over the past few decades, there has been considerable interest in the development of renovation techniques and renovation activities mainly due to the deterioration and durability problems of concrete structures. Many new techniques and a wealth of knowledge in renovation engineering have been developed and promulgated at various conferences, in papers, in press reports, and in class notes. This book aims to summarize the state-of-the-art knowledge in this field and form a systematic tool for teaching and practicing renovation activities. In the past, outmoded and functionally obsolete buildings and infrastructures were normally demolished. However, in recent years the amount of repair and refurbishment of all types of structures has increased significantly. Owners, engineers and architects of structures need to consider economic aspects as well as the historical significance and long-term serviceability by choosing either demolition and rebuilding or renovation. The owners and the public often share an ethos of conservation and adaptive reuse and their preference is usually renovation rather than demolition. Moreover, as zoning and environmental regulations make it ever more difficult to construct new buildings, renovation has become the most practical course of action. Besides, recycling buildings can be viewed as a way to convert resources and reduce landfill demand (Newman 2001). Maintaining and repairing existing building stock, and repair and replacement of the infrastructure, have been a feature of the construction industry for nearly half a century, first in Europe and then in North America. It should be noted that preparation of specification for renovation work is quite different from the design of a new structure.
2
Introduction
1.1 Building and infrastructure degradation Recently, renovation engineering has attracted increasing international attention because of the frequent occurrence of serious degradation of buildings and infrastructures. Constructed infrastructure is essential for the development and progress of commerce and industry in modern society. The gravity of infrastructure degradation can be seen from the following facts. For example, ASCE’s 2005 Report Card for America’s Infrastructure assessed the condition and capacity of US public works with an overall grade of D. ASCE estimates that US$1.6 trillion is needed over a five-year period to bring the nation’s infrastructure up to a good condition. As of 2003, 27.1 percent of the nation’s bridges (160,570) were structurally deficient or functionally obsolete; it would cost US$9.4 billion a year for 20 years to eliminate all bridge deficiencies. According to the results of a study by the Association of State Dam Safety Officials, the total investment to bring US dams into safety compliance or to remove obsolete dams tops $30 billion. About 75 percent of schools need extensive repair or replacement and the repair bill for this is as high as $268 billion according to ASCE’s 2005 Report Card for America’s Infrastructure. In 1999, the European Union set a requirement that all European highways must be able to carry 44-ton vehicles. In the UK, about 40,000 bridges cannot fulfil this requirement and need to be strengthened. Building and infrastructure degradation has become a serious social and financial problem. It can be seen that the cost for infrastructure rehabilitation has become a huge burden on the national economy of the developed countries and soon it will be the same in developing countries. Structural deterioration, together with the need to increase load-carrying capacity, has created a big market for renovation engineering. In China, according to the report of the China Academy of Engineering, the loss caused by corrosion in reinforced concrete structures reached $140 billion per year. Hence, evaluation and rehabilitation of existing infrastructures have become more and more important in the past few decades and will be more critical in the future. It is predicted that in the new century, fewer new designs and more rehabilitation work will be seen in civil engineering. More funds have to be used on inspection, maintenance, and management of existing infrastructure. More new technologies need to be developed for application in the rehabilitation of infrastructures. And, of course, there is an urgent need for a new book regarding this new branch of structural engineering.
1.2 Common causes of structural degradation It is important to understand the basic causes and mechanisms of the various forms of deterioration that degrade construction material and infrastructure made of reinforced concrete. Only with this understanding, is it possible to undertake realistic assessments of the current condition of concrete structures, and to design and implement the appropriate renovation work.
Introduction 3 Although deterioration of structure is usually a medium- to long-term process, the onset of deterioration and its rate may be influenced by the presence of defects which have their origin at the time of construction, or in the very early stages of the life of the structure (Kay 1992). Structural degradation can be divided into the following categories: (1) progressive structural failure, e.g. collapse of bridges due to repeated traffic loading and gravity loading; (2) sudden damage due to extreme loading such as fire, high speed wind or earthquake; (3) serviceability deficiencies, e.g. excessive deflections and vibrations; and (4) materials degradation, i.e. slow interaction with the environment. Deterioration of concrete can be caused by chemical attack from external sources or between the internal materials of which the structure is built, or by physical deterioration due to climatic changes, abrasion, fire, impact, explosion, earthquake, foundation failure or overloading. Specifically, the common causes responsible for structural degradations are: •
repeated loading, including: • • •
•
overloading: • •
•
•
shrinkage of constrained concrete; differential thermal expansion of layered system (e.g. asphaltic pavement on a bridge deck); expansion of internal phases (e.g. rusting steel in concrete);
severe loading or unexpected hazards: • • • • •
•
heavy materials and equipment on floors designed for light live loading; change of use resulting in higher loading than was allowed for in the original design;
non-uniform dimensional changes: • •
•
traffic loading on bridges and highways; wind-induced vibrations in bridges/buildings; machine-induced vibrations in industrial plants;
earthquake; hurricane; impact; explosion; fire which can result in some weakening of parts of the structure, as well as physical damage to columns, beams, slabs, etc.;
loss of foundation support: • •
scouring at bridge piers which may topple after loss of support; cyclic desiccation and re-hydration of clay soil;
4
Introduction •
•
abrasion/erosion of concrete surfaces: • • • • •
•
acid rain; sulfate attack; chloride diffusion;
internal chemical attack: • • • •
•
wear of pavement surface by tires of trucks; abrasion caused by steel-wheeled trolleys; abrasion of a floor slab in a factory; abrasion of marine structures by sand and shingle; erosion of hydraulic structures;
external chemical attacks: • • •
•
soil pumping under concrete pavement, with a poorly designed sub-base layer;
corrosion of reinforcing steel; alkali–aggregate reaction; stress corrosion coupled with chemical/stress effect; phase changes;
indirect effects of bacteria: •
in warm temperatures, bacteria in sewage can convert sulfur compounds into sulfuric acid. Deterioration of metallic and concrete sewage pipes can then occur.
Besides these causes due to serviceability, structural deficiency can also result from errors in design and defects in construction. It is noted that a significant proportion of the problems associated with concrete structures can be traced back to design or to construction defects (Rasheeduzzafar et al. 1989). For instance, a design consideration of inadequate concrete cover may maximize the chance of oxygen and moisture penetrating the reinforcement, thus increasing the chance of corrosion. As far as construction is concerned, one main problem is oversight of curing and this causes early age cracking that permits external agencies, such as air and moisture, to enter the concrete and attack the cement matrix and the reinforcement. Other common construction errors may include failure to place the reinforcement in the right position, and failure to provide sufficient cover for the reinforcement, or inadequate compaction for concrete. These common flaws, occurring in design, construction and serviceability, of structural degradation may cause the following defects in the structure (Chandler 1991, p. 21): •
excess deflection in beams and floors due to weak design/unforeseen loading;
Introduction 5 • • • • • • •
inadequate/insufficient fixing between precast and in-situ concrete components; lack of sufficient load-carrying packing between precast units; misalignment of precast concrete panels; inadequate movement joints between claddings and structure; inadequate insulation leading to internal condensation; surface finishes spalling or flaking; distortion of wall panels.
1.3 The objectives and scope of renovation engineering Renovation is a process of substantial repair or alteration that extends a building’s useful life (Newman 2001). Renovation engineering is a very young subject in civil engineering for concrete structures. The missions of renovation engineering are: 1 2 3 4 5
to develop a better understanding of the degradation process by identifying major parameters governing the deterioration process; to develop effective structural evaluation techniques; these techniques should be non-destructive in nature, fast and reliable; to develop economic, functional, and effective repair, strengthening, and rehabilitation techniques; to develop reliable maintenance procedures; to develop the codes and specifications for repair and rehabilitation so that public safety and health are not jeopardized.
Unfortunately, systematic studies on structural renovation of concrete are scarce and there are only a few textbooks. Other limited information has been presented only in journal papers, or special conference proceedings. So far there are no comprehensive textbooks available addressing issues on renovation of concrete structures. A confluence of several factors usually establishes the need for building renovation. Some of the common ones are: 1 2 3 4 5 6
change in use; upgrading of mechanical and electrical systems; deterioration of building envelope; structural damage and failure; upgrading of buildings for lateral loads; reducing serviceability problems.
Renovation engineering normally covers various technologies related to: (1) repair of degraded structures to recover initial load-carrying capacity; and (2) strengthening of structures to increase load-carrying capacity for current needs. The proper renovation of structures requires: (1) a good
6
Introduction
understanding of the degradation mechanisms for proper action to be taken and to avoid the recurrence of problem in the future, such as rusting when placing steel to strengthen the concrete; (2) reliable evaluation techniques for the existing condition, including a framework for structural appraisal and maintenance and non-destructive testing methods; (3) effective techniques for repair/strengthening with practical guidelines and specifications. The focus of this book will be on the rehabilitation of concrete structures, on materials and structural aspects, not on architectural features and not on utilities, and less on interactions with other engineers. On the other hand, renovation engineering is more material-oriented. Deterioration problems are basically a materials problem, especially for concrete structures. Only in the final stage of progressive failure do structural problems become significant. Renovation engineering is very practical and requires heavy field work. Inspection and field evaluation are very important in preparing the renovation work since they provide the current condition of the structure and the suggestion for remedial work for the structure. So far, not many specifications and design codes are available for renovation. Though the specific renovation work depends on the type of the structure and its condition, the following steps are generally required for a renovation job: 1 2 3 4 5 6 7 8
decision on the details of the investigation; investigation (preliminary and detailed) of the structure; diagnosis of the causes of the deterioration and evaluation of the overall condition of the structure; preparation of report to the client to suggest either renovation or rebuild; if renovation is recommended, preparation of specification and contract documents; conducting the designed renovation work; inspection of the renovation work; regular post-contract inspection and monitoring and advising on a practical program of maintenance.
1.4 Useful definitions The following common definitions are used for various terms in this book: Assessment – Systematic collection and analysis of data, evaluation, and recommendations regarding the portions of an existing structure which would be affected by its proposed use (ASCE 2000). Evaluation – The process of determining the structural adequacy or the infrastructure or component for its intended use and/or performance. Evaluation, by its nature, implies the use of personal and subjective judgment by those functioning in the capacity of experts (ASCE 2000). Infrastructure – In general, the basic economic, social, or military facilities
Introduction 7 and installations of a community, including highways, bridges, parking lots, dams and tunnels (ASCE 2000). Inspection – The activity of examining, measuring, testing, gauging, and using other procedures to ascertain the quality or state, detect errors, defects, or deterioration and otherwise appraise materials, components, systems, or environments (ASCE 2000). Rehabilitation – The process of repairing or modifying a system to a desired condition. It is an upgrade (of a damaged structure) required to meet the present needs; it implies sensitivity to building features and a sympathetic matching of original construction (Newman 2001). Repair – To replace or correct deteriorated, damaged, or faulty materials, components, or elements of a system to regain strength, density and durability. Restoration – The process of re-establishing the materials, form, and appearance of a system to those of a particular era of the system. Retrofitting – The process of increasing the load-resistance capacity or improving the performance of a structure or portion of the structure. (An example of performance improvement is to retrofit a damper into a structure to reduce its vibration.)
2
Degradation of reinforced concrete structures
Concrete is the most widely used construction material in the world, up to 10 billion tons per year worldwide consumption. Deterioration of concrete structure has become a world-wide problem and a huge burden on human society and the economy. For instance, in the UK, £500m is spent on concrete repair per year. In the USA, US$300–400m dollars are needed for the renovation of bridges and car parks alone where de-icing slats are commonly used in practice and cause severe concrete deterioration and steel corrosion. In China, the annual economic loss due to corrosion in concrete structures has reached 100 billion RMB. Deterioration is any adverse change of normal, mechanical, physical, and chemical properties either in the surface or in the body of concrete, generally due to the disintegration of its components. Degradation processes of concrete usually start from the materials level and then proceed to the structural level. They can be classified as physical (caused by natural thermal variations such as freeze–thaw cycles) artificial (such as those produced by fires), or by natural disasters (such as earthquakes and typhoon), by abrasion (erosion), chemical (attack by acids, sulfates, ammonium and magnesium ions, pure water, salts or alkali–aggregate reactions), biological (fouling, biogenic attack), and mechanical (impact, explosion, overloading, settlement, cyclic loading) (Bertolini et al. 2004). In practice, these processes may occur simultaneously, which makes things even worse. Among various degradation causes, steel corrosion is found to be the most severe problem for reinforced concrete structures that can create cracks, cause concrete cover spalling, reduce the effective crosssection area of reinforcement, and lead to collapse. The corrosion of reinforcing steel can be induced either by carbonation or chloride diffusion. Degradation of concrete occurring within the first hours to months after casting can do significant damage to mature concrete structures. As we know, cast in-situ concrete structures are hardly ever built under ideal conditions so defects may occur as the concrete is being cast or very soon afterwards, such as early-age cracking due to plastic settlement, plastic or drying shrinkage, creep, or thermal shrinkage. These defects permit the atmosphere and other environmental agencies to penetrate the body of the concrete and
Degradation of reinforced concrete structures
9
to take part in the chemical and physical processes which give rise to deterioration. There are three basic visual symptoms of deterioration in a concrete structure – cracking, spalling, and disintegration, each occurring in several different forms (Mailvaganam 1992). In a given concrete structure, the three basic indicators of deterioration may occur not only in combination, but also with several forms of each symptom being manifested simultaneously. Before any rehabilitation and renovation work can be done, the cause of the damage must be identified as clearly as possible. As far as concrete structures are concerned, degradation of reinforced concrete structures can be caused by many factors, including non-uniform dimensional changes, repeated loading, lack of durability, natural or human disasters such as typhoon, earthquake, and fire.
2.1 Degradation caused by non-uniform dimensional changes Deformation in concrete occurs mainly as the material’s response to the external load and environment. When freshly hardened concrete is exposed to the ambient temperature and humidity, it generally undergoes thermal shrinkage (shrinkage strain associated with cooling), chemical shrinkage (shrinkage associated with hydration product formation), and drying shrinkage (shrinkage strain associated with moisture loss). When the shrinkage strain is restrained, tensile stress will be built up in structural members. The degradation caused by non-uniform dimensional changes is usually associated with the difference in thermal expansion and volumetric instability of concrete. Normal concrete is very liable to dimensional changes as internal and external conditions change. This situation arises because concrete responds to both temperature and humidity effects and is almost always in a state of dynamic disequilibrium with its environments. Those dimensional changes which may be important in concrete are: (1) thermal expansion; (2) bleeding; (3) plastic shrinkage in fresh concrete; (4) drying shrinkage and cyclic swelling and shrinkage; and (5) creep. 2.1.1 Influence of non-uniform thermal expansion Non-uniform thermal expansion can be caused by the material’s different coefficients of the thermal expansion under the same heating conditions or similar materials under different thermal conditions. The coefficient of linear thermal expansion is a measure of the length change occurring in a material when it is subjected to a temperature change. Let us take a bridge deck heated by the sun as an example (Figure 2.1). During the heating process, the temperature at the top of pavement rises much faster than that at the bottom, resulting in a tendency for the pavement to bend upwards. Consequently, the concrete deck restrains the upward movement, which leads to interfacial shear stresses. However, the process is reversed during
10 Degradation of reinforced concrete structures
Figure 2.1 A bridge deck heated by the sun.
the cooling process. Due to creep, only part of the deformation is recoverable. Thus, under repeated heating/cooling processes, the pavement may finally debond from the deck and buckle. Traffic loading may also enlarge the debonded region and, eventually, the debonded and hence unsupported part of pavement cracks under traffic loading. Another example of uneven thermal expansion is thermal effects generated during the hydration process of concrete. It is well known that hydration of cement is a heat-releasing chemical reaction. The heat of hydration of cement raises the temperature of the concrete, so that the concrete is usually slightly warmer than its surroundings when it hardens, and in thick sections and with rich mixes the temperature rise may be quite considerable. For a large volume of a concrete member, the temperature distribution in the member might be quite different, higher inside and lower outside. This may cause uneven expansion or uneven dimensional changes, which could lead to cracking in concrete structures. For example, a concrete dam’s surface cools down faster than the interior. When the interior cools down, it “pulls away” from the hardened exterior surface, which may result in tensile stress, thus cracks, in different layers. As a common serviceability problem, the contraction and expansion that result due to seasonal temperature fluctuations can cause preexisting cracks to open and close. In general, thermal changes which cause damages to a structure are the rapid change when the
Degradation of reinforced concrete structures
11
concrete surface temperature changes quickly and the temperature in the core of the member changes slowly. This condition produces a curved temperature gradient with the steepest portion of the curve at the surface. As explained at the above, the upward buckling and debonding of pavement are typical examples of distress due to thermal stresses. Restrained thermal contraction is a fairly frequent cause of cracking, and often designers do not make adequate provision for thermal movements. Non-uniform thermal expansion is at first sight a material problem, but actually it is better considered as a structural problem. Temperature differences within a concrete structure result in differential dimensional changes. When the contraction or expansion is restrained, the resultant tensile stresses exceed the tensile strain capacity of the concrete and damage (cracking) can be built up. 2.1.2 Effects of bleeding Fresh concrete is a fluid mixture consisting of water, cementitious materials, sand, and coarse aggregate (gravel or crushed rock). The mixture remains plastic until the development of the cement hydrates starts to be connected each other. During concrete placing, compaction and subsequent plastic state, water or moisture tend to immigrate from bottom to top due to its smaller density. Meanwhile, aggregates tend to move downwards due to the equilibrium lost. The movements of the water or moisture will be obviously blocked by large size aggregate and reinforcing steel, resulting in a water film at the lower surface of these obstacles. Eventually, some water or moisture will be able to reach the surface of the specimen and form a layer of water film there. This phenomenon is called bleeding (see Figure 2.2). The concentration of water at the boundary of large size aggregate and reinforcing steel created by bleeding will form a weak interface in concrete. The characteristics of the interface include: large size of calcium hydroxide, less calcium silicate hydroxide, more porosity, and general weak nature. Moreover,
Figure 2.2 Phenomenon of water bleeding.
12 Degradation of reinforced concrete structures microcracks may form at the interface, sometimes even a microscopic crack. When cracks are formed in this way, their pattern on the surface tends to mirror that of the reinforcement. The water concentration on the surface of the specimen will result in more calcium hydroxides there, creating weak abrasion resistance on the surface. Water also concentrates due to its upward movement when fresh concrete is compacted. Bleeding water may be trapped at the bottom of the reinforcing bar and aggregates. In the former case, the concrete may separate from the lower surfaces of the bars and a void forms due to bleeding. If this occurs in the plastic rather than the fluid state, the concrete may crack. In such a condition, the transition zone forms after hardening and unprotected steel becomes potential site for corrosion. Due to water bleeding and sedimentation (downward movement) of coarse aggregates, paste volume increases near the concrete surface and the evaporation of water eventually slows down. The fresh concrete surface then re-absorbs water to give a higher water/cement ratio and, after hardening, resistance to surface wear and abrasion is reduced. The movements associated with the reduction in volume are resisted by the concrete which lies immediately below and which is not subject to volume change. The restraint from the lower concrete causes tensile stresses to build up in the surface layer and, because the material still remains in the fresh state or plastic and has very low strength, cracking can result. Excessive bleeding can be avoided by improving the cohesiveness of the mix through reduction of water/cement ratio, using a better aggregate grading and/or increasing the cohesiveness of fresh concrete. 2.1.3 Effects of plastic shrinkage In the fresh state, the top surfaces of concrete pours are subject to evaporation and consequent loss of the mix water. The rate of evaporation depends upon ambient conditions such as temperature, exposure to sun, wind speed and relative humidity. The water lost by evaporation is usually replaced by water rising to the surface from the bottom of the concrete by the action of bleeding. In both cases, there are local reductions in volume when the rate of removal of water from the surface exceeds the rate at which it can be replaced by bleeding. From this point of view, a concrete mixture which provides some bleeding is helpful in reducing deterioration caused by plastic cracking. Protection of concrete surface from drying winds by the use of barriers and the earliest possible application of covering to the surface may be helpful in reducing bleeding effect. ACI 305R (1999) indicates that precautions against surface drying out and cracking should be taken if evaporation is likely to exceed 1 kg m−2 h−1. Shrinkage is caused by the surface tension of water within the capillary pores formed in the cement paste during evaporation. As a capillary pore
Degradation of reinforced concrete structures
13
starts to dry out, the remaining water forms a meniscus between the adjacent cement particles, and the forces of surface tension pull the particles together. The loss of water from concrete may cause mainly two kinds of shrinkage: plastic shrinkage and drying shrinkage. Plastic shrinkage is the phenomenon when the surface layer is hardened and starts to shrink due to water loss while the inside concrete is still wet and plastic. So, plastic shrinkage is normally caused by rapid drying of fresh concrete at the surface (Allen et al. 1993). The water lost by evaporation is usually replaced by water rising to the surface of the concrete by the action of bleeding as discussed in Section 2.1.2. Where the rate of water evaporation from the surface exceeds the rate at which it can be replaced by bleeding, there is local reduction in volume. The upward bleeding of water may be accompanied by a downward movement of the solid and heavier ingredients. This downward movement may be resisted by the top layer of reinforcement or by the formwork. In the former case, the layer of concrete above the reinforcement tends to become draped over the bars. If this occurs in the plastic rather than the fluid state, the concrete may crack. In addition, the concrete may separate from the lower surfaces of the bars creating a void, as shown in Figure 2.3. When cracks are formed in this way, their pattern on the surface tends to mirror that of the reinforcement. The surface profile tends to be undulating with high points over the bars. Under other conditions, the downward movement of concrete can be restrained by the shape of the formwork. Plastic shrinkage cracks tend to propagate predominantly through the matrix rather than through the aggregate (Kay 1992). The process leading to plastic shrinkage cracking is shown in Figure 2.3. These plastic shrinkage cracks provide a path for water and other chemicals to reach the steel reinforcement, which can greatly affect the durability of concrete structures, for instance, corrosion is easily generated. In general,
Figure 2.3 Formation of plastic shrinkage crack.
14 Degradation of reinforced concrete structures if cracks appear in an exposed concrete surface very soon after it has been finished or even before finishing is complete, they are termed “plastic shrinkage cracks” (Allen et al. 1993). These cracks form when wind or heat causes the concrete to lose water rapidly, usually from 30 min. to 6h after the concrete placement. They are often wide and deep and are usually discontinuous. Plastic shrinkage cracks are usually of shallow depth, generally 38–50 mm, 300–450 mm long and usually perpendicular to the wind, which cause the water loss by evaporation. Plastic shrinkage cracks typically run parallel to one another and, because of their size, they can be structurally significant. The most effective way of preventing their occurrence is by sheltering the surface from wind and sun during construction and by covering the concrete surface immediately after finishing, measures which are all directed toward reducing the rate of evaporation. BS 8110 has recommended minimum periods of curing and protection (shown in Table 2.1) for fresh concrete to reduce plastic shrinkage. Changes in concrete mix design, and especially the use of air entrainment, may also be helpful in reducing plastic shrinkage. Remedial measures after the cracks have formed usually consist of sealing them against ingress of water by brushing in cement or low-viscosity polymers (Allen et al. 1993). The magnitude of plastic shrinkage may in extreme cases be as large as 10,000 microstrain (Troxell et al. 1968) and has been shown by L’Hermite (1960) as over 6000 microstrain for paste and 2000 microstrain for concrete. As in the plastic state, no great stress is induced in concrete and further working of the concrete can generally be applied to eliminate consequential cracks. 2.1.4 Influence of drying shrinkage The volume change and water loss of a concrete member do not stop after its initial setting and hardening. The loss of moisture is accompanied by the reduction in volume known as shrinkage. Drying shrinkage is a slow process Table 2.1 Minimum periods of curing and protection Type of cement
Portland cement, SRPC
Ambient conditions after casting
Average Poor All except Portland cement Average and SRPC, and all with Poor GGBS or PFA All Good
Min. period of curing and protection Average concrete surface temperature 5 to 10 °C
Above 10 °C
4 days 6 days 6 days 10 days
3 days 4 days 4 days 7 days
No special requirements
Degradation of reinforced concrete structures
15
in thick members, so it may lead to a gradual build-up of tensile stress if it is restrained and, further, cracking of concrete. Cracks due to restrained shrinkage are often noticed soon after construction, but when slow drying occurs they may not be apparent until much later. The drying shrinkage crack can vary from fine hairlines to as wide as 3 mm (1/8 in.), and often serve as ports of entry for moisture, carbon dioxide, and other injurious salts (Mailvaganam 1992). The formation of drying shrinkage can be explained as follows. In normal practice, in order to produce workable concrete, nearly twice as much water that is theoretically needed to hydrate the cement is usually added to the concrete mix. After concrete has been cured and begins to dry, the excessive water that has not reacted with the cement will begin to migrate from the interior of the concrete mass to the surface. As the moisture evaporates, the concrete volume shrinks. From cement chemistry, it is known that the drying shrinkage of concrete is caused by the contraction of the hydrated hardened calcium silicate gel during the withdrawal of water from the concrete. Even for hardened concrete, loss of water (e.g. during a hot season) can lead to drying shrinkage. The magnitude of the drying shrinkage is influenced by many factors, including: 1 2 3 4 5 6 7
8 9 10
the stiffness and amount of aggregate; water to binder ratio; the total amount of paste; the types and amounts of chemical admixtures; the curing regime and the age of the concrete member at which it is exposed to air; the types and amounts of mineral additives; the theoretical length of the concrete member that is defined as the ratio of section area of the member to its semi-perimeter in contact with the atmosphere; the diameter, amount and distribution of reinforcing steel; the relative humidity and its change rate; the carbonation.
Among various factors, the most important ones affecting drying shrinkage are water to binder ratio and the total amount of the paste that determines the total amount of water contained in a unit volume of the concrete. Concrete with a wetter consistency will shrink more than one with a drier consistency because the former is obtained by the use of a higher water/cement ratio, by a greater quantity of paste, or by a combination of the two. The loss of moisture from the concrete varies with distance from the surface. Drying occurs most rapidly near the surface because of the short distance the water must travel to escape and more slowly from the interior of the concrete because of the increased distance from the surface. A nearly linear relationship exists between the magnitude of the shrinkage and water
16 Degradation of reinforced concrete structures content of the mix for a particular value of relative humidity. Also, the shrinkage of concrete decreases as the relative humidity increases. When concrete is exposed to 100 percent relative humidity or submerged in water, it will actually increase in volume slightly as no water immigrates out. The gel continues to form because of the ideal conditions for hydration. If concrete is exposed to relative humidity less than 90 percent, it will normally shrink. Drying shrinkage strain, εsh, is time dependent. Approximately 90 percent of the ultimate shrinkage occurs during the first year. The magnitude of the ultimate shrinkage is primarily a function of the initial water content of the concrete and the relative humidity of the surrounding environment. For plain concrete members, drying shrinkage ranges from 400 to 700 microstrain under normal conditions. For reinforced concrete members, the shrinkage strain values are between 200 and 300 micro-strain. Because concrete adjacent to the surface of a member dries more rapidly than that in interior, shrinkage strains are initially larger near the surface than in the interior. As a result of the differential shrinkage, a set of internal selfbalancing forces, i.e. compression in the interior and tension on the outside, is built up. The stresses induced by shrinkage can be explained by imagining that the cylindrical core of a concrete cylinder is separated from its outer shell and that the two sections are then free to shrink independently in proportion to their existing water content. Since deformations must be compatible at the junction between the core and the shell, shear stresses must be created between the core and the shell. If free-body diagrams of the upper half of the cylinder are considered, it is clear that vertical equilibrium requires the shear stresses to induce compression in the core and tension in the shell. In addition to the self-balancing stresses set up by different shrinkages, the overall shrinkage creates stresses if members are restrained in the direction in which shrinkage occurs. Tensile cracking due to shrinkage will take place in any structural element restrained by its boundaries. Drying shrinkage cracks are often straight or ragged. Surface crazing on walls and slabs is an example of drying shrinkage and it is usually shallow and cosmetic in nature. Non-uniform environmental conditions produce moisture gradients which may cause differential shrinkage with resulting warping or cracking depending upon the degree of restraint experienced by the concrete – for example, curling of slabs on grade is caused by drying shrinkage and by moisture and temperature gradients. Shrinkage must be controlled since it permits the passage of water and other chemicals, is detrimental to appearance, reduces shear strength, and exposes the reinforcement to the atmosphere. For large concrete surface, joints need to be provided to prevent such cracking. Alternatively, shrinkage compensation concrete can be used to reduce drying shrinkage. The cement, used for making shrinkage compensation concrete, contains significant amounts of calcium sulfate and calcium sulfoaluminate. On hydrating, the reaction products occupy a larger volume than the original reactants. The
Degradation of reinforced concrete structures
17
volume of concrete keeps increasing during the first few days. When water loss occurs later, shrinkage occurs which reduces the volume to roughly the original value. Precautions in its application include: (1) expansive reaction stiffens up the concrete rapidly; (2) at high temperatures, above 27–29 °C, slump loss and quick setting may be a problem, so ice is normally needed to reduce the temperature; (3) quick stiffening makes plastic cracking easy and extra care is required to prevent rapid evaporation. 2.1.5 Influence of creep Creep is a continuous deformation which occurs in concrete under sustained load, and its consequences are only evident after years. Creep can thus be defined as the increase in deformation under a sustained load. The deformation increased by creep can be several times as large as the deformation under loading. Creep is thus of considerable importance in structures. Inadequate design which fails to consider the influence of creep of the structural elements of a building, for instance, shortening of columns or deflection of floors and beams, may result in load being transferred to nonstructural elements such as partition walls or cladding panels. In prestressed concrete, in flat slabs and in slender members liable to instability and buckling, creep may be harmful and the advantages of low creep concrete should be considered by the designer in these circumstances. When a sustained load is removed from a concrete member, the strain decreases immediately by an amount equal to the elastic strain at the given age, generally lower than the elastic strain on loading. This instantaneous recovery is followed by a gradual decrease in strain, called creep recovery, but the recovery of creep is not complete. Creep is not a simply reversible phenomenon, so that any sustained load results in a residual deformation. In concrete, only the hydrated cement paste undergoes creep while the aggregate is relatively hard and considered creep-free. In fact, due to its relatively high stiffness, aggregates can restrain the creep of cement paste. Therefore, creep can be considered as a nonlinear function of the volumetric content of cement paste in concrete. The volumetric content of unhydrated cement paste, the volumetric content of aggregate, the grading, maximum size, shape and texture of the aggregate and certain physical properties of aggregate will affect the magnitude of creep. Among various physical properties of concrete, the modulus of elasticity of aggregate is probably the most important factor which influences creep of concrete and the influence of other aggregate characteristics may be indirect. The higher the modulus of the aggregates, the greater the restraint provided by the aggregate to the potential creep of the hydrated cement paste. Normally aggregates with a higher porosity have a lower modulus of elasticity, thus a lower restraint to creep of concrete. Also, the porosity of aggregate plays a direct role in the transfer of the moisture within concrete, which is closely associated with creep of concrete.
18 Degradation of reinforced concrete structures As a result, some lightweight aggregates batched in a dry condition exhibit high initial creep. What’s more, the rate of creep of lightweight aggregate concrete decreases with time more slowly than in normal weight concrete. The type of cement affects the creep of concrete since cement influences the strength of the concrete at the time of application of the load. Experimental data show that within a wide range of concrete strengths, creep is inversely proportional to the strength of concrete at the time of application of load (Mehta and Monteiro 2006). Fineness of cement affects the rate of hydration and the strength development at early ages and thus influences creep. Extremely fine cements, with a specific surface up to 740 kg/m2, lead to a higher early creep but to a lower creep after one or two years under load (Bennett and Loat 1970). Strength increases in order of: low heat, ordinary, and rapid-hardening cements, so that for a constant applied stress at a fixed (early) age, creep increases in order of rapid-hardening, ordinary, and low heat cements. Creep of concrete made with expansive cement is larger than that of concrete with Portland cement only. The mechanism of creep in concrete can be related to thermally activated creep. It assumes that the time-dependent strains are the result of thermally activated processes that can be described by rate process theory. Creep strains will originate through deformation of a micro-volume of paste, called a “creep center”. The creep center will undergo deformation to a lower energy configuration under the influence of energy added to the system by external sources. This deformation can only occur by going through an energy barrier in the form of an intermediate, high-energy state. The most prevalent view involves slip between adjacent particles of C–S–H under a shear stress. If there is a sufficient amount of water between layered C–S–H, which reduces the van der Waals’ forces sufficiently, slippage is ready to occur and hence the creep. Creep can also result from the diffusion of micro water under stress (Mindess et al. 2003). Water-reducing and set-retarding admixtures lead to pore refinement in the hydration product and have been found to increase the basic creep in many, but not all cases (Hope et al. 1967; Jessop et al. 1967). The environmental humidity of concrete is also an important factor influencing creep. An increase in the atmospheric humidity is expected to slow down the relative rate of moisture flow from the interior to the outer surfaces of concrete. Taking a broad view, for a given concrete, the lower the relative humidity, the higher the creep. The strength of concrete has a significant influence on creep: within a wide range, creep is inversely proportional to the strength of concrete at the time of application of the load. The size and the shape of a concrete element also have some influence on the magnitude of creep since the rate of water loss is obviously controlled by the length of the path travel by the water and the resistance of water escaping from the interior of the concrete is closely related to creep. Other factors influencing creep include the curing history of concrete, the temperature of exposure and the applied stress. Creep strains can be
Degradation of reinforced concrete structures
19
significantly different when concrete elements are cured in different histories. For instance, drying cycles can enhance micro-cracking in the transition zone and thus increase creep. The exposure temperature of concrete can have two counteracting effects on creep. On the one hand, if a concrete member is exposed to a higher than normal temperature as part of the curing process before it is loaded, the strength will increase and the creep strain would be less than that of a corresponding concrete stored at a lower temperature. On the other hand, exposure to high temperature during the period under load can increase creep. As far as the applied stress is concerned, there is a direct proportionality between creep and the applied stress for hardened concrete. It appears safe to conclude that, within the range of stresses in structures in service, the proportionality between creep and stress holds good, and creep expressions assume this to be the case. Also, creep recovery is also proportional to the stress previously applied (Yue and Taerwe 1992). Creep is usually determined by measuring the change with time in the strain of a specimen subjected to a constant stress and stored under appropriate conditions. ASTM C 512-94 describes a spring-loaded frame to measure the creep of a concrete sample which maintains a constant load on a concrete test cylinder despite any change in its length. Since creep may develop over as long as 30 years, the measurement can only cover a short period of the age of concrete. Numerous mathematical expressions relating creep and time have been suggested, among which includes the modified Ross expression of ACI 209R-92 (1994a) and those suggested by Bazant and his co-workers (1992). Creep affects strain, deflection and, often, stress distribution. In concrete structures, creep reduces internal stresses due to non-uniform shrinkage, so that there is a reduction in cracking (Neville et al. 1983). In reinforced concrete columns, creep results in a gradual transfer of load from the concrete to the reinforcement. In an eccentrically loaded column, creep increases the deflection and can lead to buckling. In statically indeterminate structures, creep may relieve stress concentrations induced by shrinkage, temperature changes, and so on. However, in mass concrete, creep may cause cracking when a restrained concrete mass undergoes cyclic temperature changes. Creep can also lead to an excessive deflection of structural members and cause other serviceability problems, especially in high-rise buildings and long bridges.
2.2 Degradation caused by repeated loading Components of reinforced concrete structures such as machine foundations and bridges are frequently subjected to repeated loading (cyclic loads), and the resulting cyclic stresses can lead to microscopic physical damage in the materials. The damage can accumulate and further lead to the strength reduction and then structural degradation. The trend of strength reduction
20 Degradation of reinforced concrete structures
Figure 2.4 Normalized fatigue strength as a function of number of loading cycles.
can be seen from the so-called S–N curve as shown in Figure 2.4. The cyclic fatigue can be defined as a failure caused by a sufficient repeated application of loads that are not large enough to cause failure in a static application. This implies that some internal progressive permanent structural damage must be accumulated in the reinforced concrete structure under the repeated stress. A typical cyclic loading is shown in Figure 2.5. Some useful definitions and basic concepts for cyclic loadings can be introduced, referring to Figure 2.5: Constant amplitude stressing – cycling between maximum and minimum stress levels that are constant (see Figure 2.5(a)); Stress range, Δσ, is the difference between the maximum and the minimum values; Δσ = σmax − σmin
(2.1)
Mean stress, σm, is the average of the maximum and minimum values; σm =
σmax + σmin 2
(2.2)
Stress amplitude, σa, is the half of stress range; σa =
Δσ 2
=
σmax − σmin 2
(2.3)
Completely reversed stressing – mean stress equal to zero with constant amplitude; Stress ratio, R, is σmin /σmax; Amplitude ratio, A, is σa /σm
Degradation of reinforced concrete structures
21
Figure 2.5 Different types of cyclic loadings.
In cyclic fatigue, the symbol “S” is usually used to represent nominal or average stress, which is different from the true stress at a point, σ. Nominal stress distribution is determined from the load or the moment using formulae in mechanics of materials as a matter of convenience while true stress is determined according to the real materials state (stress concentration, yielding). S is only equal to σ in certain situations. The fatigue strength of a material is largely influenced by the maximum stress applied, the difference between maximum and minimum stress (stress range), and the number of cyclic loading. It should be noted that for one cycle of loading and unloading, the concrete stress–strain curve is a closed cycle (see Figure 2.6). The area enclosed is proportional to hysteresis and represents the irreversible
22 Degradation of reinforced concrete structures
Figure 2.6 Hysteresis loop during loading and unloading.
energy of deformation, i.e. energy due to crack formation or irreversible creep. The fatigue life of a material is usually plotted as a figure of nominal stress versus cyclic number, S–N diagram. To get an S–N diagram, fatigue tests have to be conducted. Each test deals with a fixed stress amplitude and mean stress. The test continues until the specimen fails at a cyclic number of N. Each experiment result will generate one point on the S–N diagram. The S– N diagram should have sufficient data points to make the empirical analysis meaningful. To make things simple, usually a completely reversed stressing, mean stress equalling to zero with constant amplitude, is adopted first to build up the S–N diagram. For the cases of mean stress not equalling zero, fatigue life can be estimated by using the S–N diagram of completely reversed stressing as stated in the following section. When sufficient experimental data are obtained from completely reversed stressing fatigue test, the S–N diagram can be plotted in a linear-linear coordinates, or linear-log coordinates, or log-log coordinates. If S–N data are found to be a straight line on a log-log plot, the relationship between stress amplitude and fatigue cycles can be written as: σar = A(Nf )B
(2.4)
where σar is the stress amplitude for completely reversed stressing corresponding to Nf. The A and B are material constants; and Nf is the cycle to failure. For the cases where σm ≠ 0, the relationship between the stress amplitude where σm ≠ 0 and the stress amplitude for completely reversed stressing can be expressed by the empirical modified Goodman law,
冢
σa = σar 1 −
σm σµ
冣
(2.5)
Degradation of reinforced concrete structures
23
where σa is the stress amplitude that σm ≠ 0 for a given fatigue life, σar the stress amplitude of completely reversed stressing at fatigue failure (Nf), σm the mean stress, and σµ is the static strength of the material. This equation provides the base for estimating the fatigue life for a case that σm ≠ 0 by utilizing the S–N diagram for completely reversed stressing. Goodman’s law can also be given in the form of stress range. To keep the same number of cycles to failure: Δσ = Δσr(1 − σm/σu)
(2.6)
where Δσ is the stress range when σm ≠ 0; and Δσr is the stress range when mean stress = 0. The fatigue life of any (σm, σa) combination can be estimated from the following procedures. First, substitute Eq. (2.4) into (2.5), we get:
冢
σa = 1 −
σm σµ
冣 AN
B f
(2.7)
It can be seen that this equation reduces to σa = A(Nf)B as it should, if σm = 0. On a log–log plot, Eq. (2.7) produces a family of S–N curves for different values of mean stress, which are all parallel straight lines. In general, let the S–N curve for completely reversed loading be
冢
σa = 1 −
σm f(Nf ) σµ
冣
(2.8)
which is the corresponding family of S–N curves. From Eq. (2.7), we can get:
冪A(σ − σ )
Nf = B
σa
µ
(2.9)
m
If more than one amplitude or mean level occurs in a fatigue test, fatigue life may be estimated by summing the cycle ratios, called the Palmgren-Miner Rule. Σ(Ni /Nfi ) = 1
(2.10)
where Ni is the number of applied cycles under Δσi or σai and Nfi the number of cycles to failure under Δσi or σai. Often, a sequence of variable amplitude loading is repeated a number of times. Under these circumstances, it is convenient to sum cycle ratios over one repetition of the history, and then multiply this by the number of repetitions required for the summation to reach unity.
24
Degradation of reinforced concrete structures Bf
冤冱N冫N 冥 i
fi
=1
(2.11)
one repetition
where Bf is the number of repetitions to failure. Another approach for fatigue life prediction involves fracture mechanics concepts. Consider a growing crack that increases its length by an amount Δa due to the application of a number of cycles ΔN. The rate of growth with cycles can be characterized by da/dN. Assume that the applied loading is cyclical with constant values of the loads Pmax and Pmin, hence also with constant values of the nominal stresses Smax and Smin. For fatigue crack growth, it is conventional to use the nominal stresses that are generally defined based on gross area to avoid the change of stress values with crack length. The primary variable affecting the growth rate of a crack is the range of the stress intensity factor. This is calculated using the stress range ΔS: ΔK = FΔS冪πa
(2.12)
The value of F depends only on the geometry and the relative crack length, α = a/b just as if the loading was not cyclic. Since K and S are proportional for a given crack length, the maximum, the minimum, and the range for K during a loading cycle are given by Kmax = FSmax冪πa
(2.13)
Kmin = FSmin冪πa
(2.14)
ΔK = Kmax − Kmin
(2.15)
For a given material and set of test conditions, the crack growth behavior can be described by the relationship between cyclic crack growth rate da/dN and stress intensity range K. The empirical fitting curve suggests that the following relationship can be used, da dN
= C(ΔK)m
(2.16)
where C and m are curve fitted constant (from log–log plot). It should be indicated that Eq. (2.16) is obtained empirically and valid for an intermediate crack growing rate or ΔK range. At low growth rates, the curve generally becomes steep and appears to approach a vertical asymptote denoted Kth, which is called the fatigue crack growth threshold. This quantity is interpreted as a lower limiting value of K, below which crack growth does not ordinarily occur. At high growth rates, the curve may again become steep. This is due to rapid unstable crack growth just prior to the final failure of the test specimen.
Degradation of reinforced concrete structures
25
The number of cycles to failure during a fatigue test can be calculated using the following equation: Nf =
冮
af
ai
da C[ΔK]m
(2.17)
where ai = initial crack size obtained from inspection (if no crack is found, take ai as the crack detection threshold); and af = final crack size obtained from Kmax(af) = Kc. By substituting the parameter expressions in Eq. (2.17), we can obtain: ai
⎡ 1 − 冢a 冣 冢 2 冣 ⎤ 1 f ⎥ m Nif = ⎢ ⎢ C(FΔS冪π)m m − 1 ⎥ ai 冢 2 − 1 冣 冢2 冣⎦ ⎣ m
−1
(2.18)
where Nif is the cyclic number for material fail in fatigue from crack growing from ai to af ; F is a geometrical function that depends on loading pattern and ratio of crack size to specimen size.
2.3 Degradation caused by lack of durability Almost universally, concrete has been specified principally on the basis of its compressive strength at 28 days after casting. In many cases, the specified strength can be achieved with a high water/cement ratio and hence permeable concrete. R.C structures, on the other hand, are almost always designed with a sufficiently high safety factor. Thus, it is rare for concrete structures to fail due to lack of intrinsic strength. However, gradual deterioration caused by the lack of durability makes reinforced concrete infrastructures fail to survive their specified service lives in ever increasing numbers. The extent of the problem is such that concrete durability has been described as a “multimillion dollar opportunity” (Anonymous 1988). In a report by the National Materials Advisory Board of the National Research Council, USA, it suggests that there is a lack of proper application of the durability knowledgebase by practitioners, as evidenced by the continuing problem of inadequate durability of concrete in service. The report also notes that the magnitude of durability problem is great enough to merit national action, and seeks reasons for the lack of proper implementation of available data by engineers. It finds that available concrete technology is inadequately used because of the fragmentation of the industry, confused responsibility for the training and skill of the workers, lack of financial or other incentives for the industry, and poor management and dissemination of available technical information. It further finds that two reasons may contribute to the poor durability of
26 Degradation of reinforced concrete structures national infrastructure. The first is that the contractors, who can most directly provide it, do not make durability their responsibility. The second is a product of economic management and tax structure. Most developers have no intention of maintaining the ownership of new buildings for the life of those buildings because it is advantageous to sell them within a few years of construction. The initial owner has therefore little incentive to spend money to ensure continued durability when the benefits will not be realized during the period of his ownership. The report argues that for many construction projects neither the owner nor the contractor is motivated to ensure longterm durability of the structures, yet there is potentially a severe financial loss to the community. Restoration of durability is by far the most common repair work of concrete structures. According to ACI Committee 201, the durability of Portland cement concrete is defined as its ability to resist weathering action, chemical attack, abrasion, or any other process of deterioration to keep its original form, quality and serviceability when exposed to its intended service environment (Mehta and Monteiro 2006). Durability is most likely to relate to longterm serviceability of concrete and concrete structures. Serviceability refers to the capability of the structure to perform the functions for which it has been designed and constructed with exposure to a specific environment. The structure should be able to resist or withstand, during its service life, all the intended loads and environmental conditions without excessive deterioration, wear or failure. Thus, durability in the broadest sense will depend on the nature of the concrete and the aggressiveness of the in-service environment. Durability of concrete is very dependent on the proportioning of the concrete mix. Good concrete quality and adequate reinforcement cover are important factors for the durability of concrete structures. Permeability is another key factor in durable concrete and concrete structures. Concrete permeability is governed by the water/cement ratio, the cement content, the curing and the degree of compaction. Overbatching of water or under-batching of cement are construction errors which can lead to problems early on, particularly in aggressive environments, because they cause more pores in the matrix and it is thus more permeable. The requirement for the reinforcement cover is usually specified with respect to exposure conditions that the concrete is supposed to serve. Normally design codes have reasonable specifications for the minimal cover thickness, which is important to prevent corrosion of reinforcement and ensure durability of concrete structures. It is now known that, for many conditions of exposure of concrete structures, both strength and durability have to be considered explicitly at the design stage. Therefore, the provision of durability consists mainly of a prescription of maintaining the required margins and factors with time (Keyser 1980). There are four main methods that ensure adequate durability of concrete in service as follows:
Degradation of reinforced concrete structures 1 2 3 4
27
compliance with current standards of good practice during construction; the use of new and improved materials and innovative construction systems designed for increased durability at competitive cost; provision of protection to existing undamaged structures against adverse environments; use of materials and procedures that incorporate the best available standards of good practice in repair, replacement, and subsequent protection of already damaged structures.
2.3.1 Causes of deterioration and main durability problems The causes of degradation of Portland cement concrete can be classified as different groups of factors that can affect the performance of a building material, component, or system. For the sake of clarity, the classification of causes of concrete deterioration can be roughly grouped into three categories in this book: (1) physical causes; (2) chemical causes; and (3) mechanical causes. Physical causes of deterioration and durability problems include the effects of high temperature or of the differences in thermal expansion of aggregate and of the hardened cement paste, such as alternating freezingthawing cycle and the associated action of de-icing salts, surface wear or loss of mass due to abrasion, erosion, and cavitation, and cracking which appear widely due to volume changes, normal temperature and humidity gradient, crystallization of salts in pores, structural loading, restrained shrinkage, and exposure to fire. The most common chemical causes of problems with concrete durability are: (1) hydrolysis of the cement paste component; (2) carbonation; (3) action-exchange reaction; and/or (4) reaction leading to expansion (such as sulfate expansion, alkali–aggregate expansion, and steel corrosion) (Mehta and Gerwick 1982; Mehta and Monteiro 2006). Chemical degradation is usually the result of an attack, either internal or external, on the cement matrix. Portland cement is alkaline, so it will react with acids in the presence of moisture and, in consequence, the matrix may become weakened and its constituents may be leached out. Mechanical causes include impact and overloading. In reality, major durability problems of concrete structures include corrosion of the reinforcing steel, freeze/thaw damage, salt scaling, alkali–aggregate reactions, and sulfate attack. The common result of these attacks is that all of them can result in cracking and spalling of the concrete. It should be noted that the physical and chemical processes of deterioration usually act in a synergistic manner and the deterioration and durability problem of concrete is rarely due to one isolated cause.
28 Degradation of reinforced concrete structures 2.3.2 Basic factors influencing durability Concrete is a permeable and a porous material. The capillary pore structure allows water under pressure to pass slowly or ions or moisture under concentration gradient to migrate gradually through concrete. The property that governs the rate of flow of a fluid into a porous solid under pressure is defined as permeability. The property that governs the rate of migration of ions in a porous solid under concentration gradient is defined as diffusivity. The durability of concrete depends, to a large extent, on permeability and diffusivity. In fact, with the exception of mechanical damage, all the adverse influences on durability involve the transport of fluids through the concrete, which is closely related to concrete permeability and diffusivity. From a microstructure point of view, permeability is greatly affected by the nature of the pores, both their size and the extent to which they are interconnected. The average size of capillary pores in concrete is about 0.1 µm while the gel pores are very much smaller. In general, concrete permeability is affected by: (1) the quality of cement and aggregate; (2) the quality and quantity of the cement paste, including the amount of cement in the mixture, water/cement ratio and degree of hydration of cement; (3) the bond developed between cement paste and aggregate; (4) the effectiveness of compaction of the concrete; (5) the extent of curing; (6) the presence or absence of cracks; and (7) the characteristics of admixtures used in the concrete. Normally aggregates also contain pores, but these are usually discontinuous. Moreover, aggregate particles are enveloped by the cement paste so that the pores in the aggregate do not contribute to the permeability of concrete (Neville 1996). For steady-state, the coefficient of permeability (K1) can be determined by Darcy’s law that describes a volume flow rate of water under the equilibrium flow condition HA dq = K1 dt Lµ
(2.19)
where dq/dt is the volume flow rate (m3 s−1); H is the pressure head loss across the thickness of the medium (m); A is the surface area of the medium (concrete) normal to the direction of flow (m2); L is the thickness of the medium (m); µ is viscosity of the fluid; and K1 is the coefficient of permeability depending on the properties of the medium and of the fluid, normally water in the case of concrete, (m s−1). Obviously the permeability is a function of the pores inside materials. This includes two concepts, percentage of porosity and size distribution of pores. In comparison, porosity is a measure of the proportion of the total volume of concrete occupied by pores and it determines the quantity of liquid or gas that can be contained by a concrete. The size distribution of the pores decides the inter-connecting properties of pores. If the pores are interconnected, the transport of fluids through concrete is easier, leading to high
Degradation of reinforced concrete structures
29
permeability. On the other hand, if the pores are discontinuous or ineffective with respect to fluid transport, the permeability of the concrete is low, even if its porosity is high. The permeability of naturally hardened paste, that had never been allowed to dry, was found by Powers et al. (1954) in a range from 0.001 × 10−12 to 1.20 × 10−12 m s−1 for water/cement ratios ranging from 0.3 to 0.7. Drying was found to increase the permeability. As cement hydration proceeded, the permeability was significantly reduced and in 24 days was found to be only one-millionth of its initial value. Table 2.2, after Powers et al. (1954), shows this effect. For cement containing higher proportions of C3S, the time needed to produce complete closure of the capillary pores can be less, leading to lower permeability at an early age. Since water flows more easily through the capillary pores than through the much smaller gel pores, the permeability of the cement paste as a whole is 20–100 times greater than that of the gel itself (Neville 1996). Different from permeation that refers to flow under pressure differential, diffusion is the process in which a medium moves under a differential in concentration and the relevant property of concrete is referred to as diffusivity. In other words, diffusivity is the rate of moisture migration at the equilibrium diffusion condition and it is defined by Fick’s law. Fick’s first law is: J = −D
dc
(2.20)
dx
where J is the diffusion flux or mass transport rate per unit area per unit time of a medium in kg/m2s. C is concentration in kg/m3, D is diffusion coefficient in m2/s, x coordinate or thickness of the sample in diffusion direction, and dC/dx is concentration gradient in kg/m4. Even though diffusion takes place only through the pores, the values of J and D refer to the cross-section of the concrete sample; thus, D is actually the effective diffusion coefficient. Fick’s second law is: ∂C ∂t
=D
∂2C ∂x2
(2.21)
Table 2.2 Reduction of permeability of cement paste (water/cement = 0.7) with the progress of hydration Age (days)
Permeability coefficient (× 10−12 ms−1)
Fresh 5 6 8 13 24 Ultimate
2,000,000 400 100 40 5 1 0.6
Source: After Powers et al. (1954).
30 Degradation of reinforced concrete structures The difference between the two parameters, K1 and D, is that permeability is the parameter characterizing water flow when the pores inside the concrete are filled with water, while diffusivity is the parameter describing the diffusion of water vapor before saturation is reached in the pores. K is a function of pressure difference and D is concentration difference. If one of the parameters is known, the other can be deduced indirectly. As far as the diffusion of gases in concrete is concerned, carbon dioxide and oxygen are of primary importance: the former leads to carbonation of hydrated cement paste, and the latter makes possible the progress of corrosion of embedded reinforcement steel. Like permeability, diffusion is low at lower water/cement ratios, but the influence of the water/cement ratio on diffusion is much smaller than on permeability. 2.3.2.1 Permeability test Permeability tests measure the rate at which a liquid or gas passes right through the test specimen under an applied pressure head. Concrete is a kind of porous material which allows water under pressure to pass slowly through the concrete, but the rate of flow through dense, good quality concrete is extremely difficult. Till now, testing the permeability of concrete has not been generally standardized (Ludirdja et al. 1989). There are two common practices to evaluate the permeability of concrete using water as permeation liquid: the steady flow method and the depth of penetration method. The steady flow method is performed on a saturated specimen and applies a pressure head to one end of the sample. When a steady flow condition is reached, the measurement of the outflow enables the determination of the coefficient of permeability by using Darcy’s law: dq
L dt k1 = HA
(2.22)
where: k1 = coefficient of permeability (m/sec) dq/dt = rate of the steady flow (m3/sec) L = thickness or length of the specimen (m) H = drop in hydraulic head across the sample (m) A = cross-sectional area of the sample (m2). To evaluate the coefficient of permeability by the steady flow method, water should be absorbed into all pores of the sample so that the pore surfaces do not provide friction nor capillary attraction to the passage of water. However, such flow conditions cannot always be achieved in many low
Degradation of reinforced concrete structures
31
permeability concretes and are not representative of an actual working environment. Moreover, the considerable length of time required to test the concretes, and the difficulties of attaining a steady state outflow, can be regarded as disadvantages for the steady flow method. In cases of good quality concrete, there is no flow of water through the concrete, making the steady flow method not suitable for measuring the permeability of concrete. Under such conditions, the depth of penetration method is a good choice for measuring the permeability of concrete. In the penetration method, one end of the unsaturated concrete specimen is subjected to a pressure head while the other end is free in normal atmospheric conditions. The measure of water penetration is achieved either by measuring the volume of water entering the sample or by splitting the cylinder and measuring the average depth of discoloration, due to wetting, taken as equal to the depth of penetration. Provided the flow of water is uniaxial, the water penetration depth can be approximated to the coefficient of permeability equivalent to that used in Darcy’s law as developed by Valenta (1969): k=
x2v 2hT
(2.23)
where: x = depth of penetration of concrete (m) v = the fraction of the volume of concrete occupied by pores h = hydraulic head (m) T = time under pressure (s). The value of v represents discrete pores, such as air voids, which are not filled with water except under pressure, and can be calculated from the increase in the mass of concrete during the penetration test. Bearing in mind that only the voids in the part of the specimen penetrated by water would be considered, we can write: v=
ΔW ρAx
(2.24)
in which ΔW is the gain in weight of the specimen during the penetration test, ρ is the density of concrete, A is the area, and x is the penetration depth. Typically the effective porosity, v, lies between 0.02 and 0.06 (Vuorinen 1985). Based on the depth of penetration method, it is possible to use the depth of penetration of water as a qualitative assessment of concrete permeability: a penetration depth of less than 50 mm classifies the concrete
32 Degradation of reinforced concrete structures as “impermeable”; a depth of less than 30 mm as “impermeable under aggressive conditions” (Neville 1996). It is now acknowledged that the steady flow method suits concretes with relatively higher permeability, while the depth of penetration method is most appropriate for concrete with very low permeability. It is important to note that the scatter of permeability test results on similar concrete at the same age, and using the same equipment, is large. For any set of tests, the value of K1 in Eq. (2.19) depends on both the medium and the fluid and, therefore, represents the permeability of the medium at a specified temperature. 2.3.2.2 Chloride diffusion test There are basically two methods in evaluating the chloride diffusion in concrete. One method is called the rapid permeability test (ASTM C1556-04) and another is the diffusion cell test method (Li et al. 1999). The latter is considered a reliable test method for the chloride diffusivity of concrete due to the difference of concentration. In this method, the specimens of Φ100 × 20 mm (Φ3.94 × 0.79 in.) slice are placed between two chambers and the edges are sealed with an epoxy resin (see Figure 2.7). After the epoxy resin is cured, saturated calcium hydroxide solution is poured into chambers and the specimens are immersed in the solution for five days. This procedure is to avoid an anomalous effect due to sorption rather than diffusion of chloride ions. Then a NaCl solution with a concentration of 5 Mol is added into the Chamber A to start the chloride diffusion test. The chambers are maintained at 23 ± 2°C and the concentration of the chloride diffused through the specimens in Chamber B is measured periodically. The testing process is as follows: • • • • • • •
Prepare the solution with a prepared concentration of 0.1 Mol, 0.01 Mol and 0.001 Mol sodium chloride in saturated calcium hydroxide in advance. To check the chamber with high concentration of sodium chloride, use the 0.1 Mol and 0.01 Mol solution for calibration. After calibrating the meter, withdraw 1 ml aliquot from the chamber tested. Dilute it into 100 ml with saturated calcium hydroxide solution and add 2 ml of sodium nitrite (5 Mol concentration) to the solution to increase the sensitivity of the instrument. Put the chloride ion selective electrode into the solution and record the data when the reading of the meter becomes stable. Add the appropriate amount of sodium chloride to maintain 5 Mol concentration. To test the chamber with a low concentration of sodium chloride, use 0.01 Mol and 0.001 Mol solution for calibration.
Degradation of reinforced concrete structures
33
Figure 2.7 Chloride diffusion test.
• • •
Withdraw 10 ml aliquot from the solution in the chamber, and add 0.2 ml of sodium nitrite (5 Mol concentration) into the solution. Put the chloride ion selective electrode into the solution and record the reading when the meter becomes stable. Refill the chamber with 10 ml solution of appropriately prepared concentration.
A typical curve of chloride concentration in Chamber B obtained from the experiment has a strong nonlinearity between chloride concentration and time at beginning (see Figure 2.8). However, when the test time exceeds about 30 weeks, the curve becomes quite linear. This implies that the chloride diffusion reaches a steady state. The linear relation between concentration of chloride ions and time can be expressed as: C = kt − A
(2.25)
34 Degradation of reinforced concrete structures
Figure 2.8 Diffusion test results.
where C is the cumulative concentration of chloride ion diffused into Chamber B at time t and k is the slope of chloride concentration-time curve in steady state. A is a constant. On the other hand, in steady state, the flux of chloride ions through the specimens can be described by Fick’s first law of diffusion. For a case of steady state, this law can be written in the following form: C=
DCA l2
t−A
(2.26)
where D is the diffusion coefficient in steady state; CA is the concentration of chloride ion in Chamber A; and l is the thickness of the slice specimen. Comparing the above two equations, the diffusion coefficient can be expressed as follows: D=
kl 2 CA
(2.27)
It is clear that once the slope of chloride concentration–time curve (flux) in the steady state is known, the diffusion coefficient can be calculated. In some instances, additives are introduced into concrete to accelerate the curing. There are two reasons for this, first, to achieve a quicker turn round of molds or formwork, and, second, to allow concreting to take place in cold weather. A constituent of these additives is calcium chloride (CaCl2). There is still some doubt as to the effect of calcium chloride on reinforced concrete, but evidence is emerging that it can have a disastrous effect by corroding the rebars. Chloride ions in concrete are not always distributed uniformly and
Degradation of reinforced concrete structures
35
their concentration has resulted in severe corrosion of rebars in structural members, especially where they are exposed to the weather. 2.3.3 Deterioration due to physical action Progress loss of mass from a concrete surface can occur due to surface wear. Concrete surfaces can be produced to have a high degree of resistance to surface wear, including abrasion, erosion and cavitation. The resistance of a concrete surface is largely dependent on achieving a hard and durable surface which is flat and free from cracks. Abrasion of a concrete surface can be defined as the worn process by dry attrition, repeated rubbing, rolling, sliding, or frictional (attrition) processes (Mehta 1980). Surface abrasion is mainly caused by dry attrition. Pavements and industry floors are most likely susceptible to dry attrition, so are hydraulic structures. Two categories of abrasion are possible for concrete roads: mechanical abrasion (such as attrition, scraping and impact) due to heavy trucking and vehicles with chains or studded tires; and polishing of the surface due to traffic action resulting in loss of skid resistance. In hydraulic structures, the abrasive action results from the cutting action of suspended solids. Concrete abrasion resistance depends mainly on: (1) concrete strength; (2) cement content; (3) water/cement ratio; (4) aggregate properties; (5) finishing methods; (6) good compaction and curing; and (7) use of topping. The abrasion resistance of concrete varies in proportion to both the compressive strength and cement contents and inverse in proportion to the water/cement ratio regardless of aggregate quality (Mailvaganam 1992). Concrete strength has a significant influence on the abrasion resistance of concrete surface. The finishing method applied to the surface is also important, but its influence on abrasion is difficult to be quantitatively evaluated. Aggregate property also plays an important role in concrete resistance to abrasion. Concrete containing harder coarse aggregate is normally more resistant to abrasion than that containing softer aggregate. To increase resistance to surface abrasion, the finishing procedure should be done to maintain as much coarse aggregate near the top surface as possible. Good abrasion resistance concrete, such as micro silica (MS) concrete and densified with small particles (DSP) concrete, may be used for concrete structures that are most liable to be subject to dry attrition. Under conditions of severe wear, aggregates featuring high resistance to abrasion should be selected. Commonly such materials include: (1) metallic types, such as pearlitic iron turnings and crushed cast iron chilled grit; and (2) non-metallic types, such as silicon carbide grains (carborundum); fused alumina grains (corundum); and natural emery grains (alumina and magnetite) (Campbell-Allen and Roper 1991). In cases where slip resistance is very important, or where rusting of the metals would be objectionable, the non-metallic types are preferred. Fiber reinforced concrete has also been demonstrated to be much superior in wear resistance to conventional concrete (Neville 1975).
36 Degradation of reinforced concrete structures Penetrating sealing and hardening treatments can significantly increase abrasion resistance (Sadegzadeh and Kettle 1988). Erosion refers to concrete surface wear by the abrasive action of fluids with suspended solid particles. Erosion occurs in hydraulic structures, for instance, canal linings, spillways, piers and concrete pipes for water or sewage transport. When a fluid containing suspended solid particles is in contact with concrete, the impinging, sliding, or rolling action of the particles will cause surface wear. The rate of surface erosion depends on the porosity or the strength of concrete, and on the amount, size, shape, density, hardness, and velocity of the moving particles. Better resistance to erosion is achieved by using hard aggregate; MS concrete and DSP concrete; and concrete of strength greater than 41 MPa cured at least for 7 days. Concrete surface quality has a significant influence on resistance to surface wear. The properties of the concrete surface layer are determined largely by the quality of the concreting operations and the timing of these operations, particularly the timing of finishing and curing. To improve surface quality, it is suggested to do surface finishing right before the initial set and/or to add topping with low water/cement ratio and hard aggregates. Hydraulic structures may be intensively subjected to physical erosion which arises from particles of rock, ice or debris carried in the flowing water; from the collapse of vapor bubbles formed by pressure changes in highvelocity flows; and from the fluctuations of water pressure in and on the concrete under conditions of unsteady flow (Campbell-Allen and Roper 1991). These vapor bubbles flow downstream and, on entering a zone of higher pressure, collapse with great impact. Repeated collapse of such cavities near the surface of concrete will cause pitting. Erosion of hydraulic structures easily leads to cavitation, which relates to loss of mass by the formation of vapor bubbles and their subsequent collapse due to a sudden change of direction in rapid flowing water. Cavitation erosion is readily recognized from the nature of the holes or pits formed, which are quite different from the smoother worn surface produced by abrasion from solids. In hydraulic structures, once cavitation damage has started, the roughened surface provides a source for new cavities to form and the damage can be extended far downstream. Advanced stages of the damage show an extremely rough honeycomb texture with some holes that penetrate deep into the concrete. The erosion of concrete by sand and gravel in water can be equally as severe as that caused by cavitation. Erosion progresses rapidly after an initial roughening of the surface occurs. The material beneath the surface is more vulnerable to attack because of the tendency of the erosion to follow the mortar matrix and undermine the aggregate. Better resistance to erosion is achieved by using high strength concrete containing the maximum amount of hard coarse aggregate and/or very strong concretes having strengths over 100 MPa with high-range water-reducing admixtures and silica fume (Holland 1983).
Degradation of reinforced concrete structures
37
Better resistance to cavitation in hydraulic structures is achieved by using fiber reinforced concrete to increase the lifetime of hydraulic structures. For hydraulic structures, damage can also be initiated by chemical attack in addition to the strictly mechanical type of damage that results from wear and erosion. In such cases, things can be even worse. Mineral water with an acidic pH can cause the leaching of lime or can etch the surface of concrete. The weakened concrete and the roughened surfaces are then more vulnerable to the intense attrition caused by the forces of erosion. This kind of acid attack produces the disintegration of the concrete which is then eroded away. Impact by high speed water can also damage concrete. Nonlinear flow at velocities exceeding 12 m/s may cause severe damage to concrete surface through cavitation. The high velocity water jet can cause cavitation to form on concrete surface with irregularities. The impact of a very high velocity jet of water can strike the concrete surface and erode the cement paste, resulting in the boring of a hole through the concrete. Useful measures to improve concrete cavitation resistance, caused by high speed water jet, can be the removal of surface misalignments or abrupt changes of slope. 2.3.4 Degradation caused by steel corrosion It is thought that a high percentage of the defects and deterioration in concrete structures arises from the corrosion of the reinforcing steel. Originally, in a properly designed, constructed and used structure, there should be little problem of steel corrosion within the concrete during the design life of that structure because that concrete provides a very high alkali environment with Ph value of 12.5–13.5 in which steel is well passivated. Unfortunately, this highly desirable condition is not always found during the service period of concrete structures, resulting in that corrosion of reinforcement has become probably the most frequent cause of damage to reinforced concrete structures. The processes of the degradation of concrete and the corrosion of reinforcing steel are closely related. The former provokes destruction of the concrete cover and causes microcracking that compromises its protective characteristics. On the other hand, corrosion attack, because of the expansive action of corrosion products, generates cracking or delamination of the concrete and reduces the bonding between the reinforcement and surrounding concrete. Based on these considerations, the service life of reinforced-concrete structures can be divided into two distinct phases: the initiation of corrosion and the propagation of corrosion (Bentur et al. 1997; Bazant 1979). During the initial phase, aggressive substances, such as carbon dioxide and chlorides, can penetrate from the surface into the bulk of the concrete and depassivate the steel. The duration of the initiation phase depends on the cover depth and the penetration rate of the aggressive agents as well as on the concentration necessary to depassivate the steel. Once the protective layer on
38 Degradation of reinforced concrete structures reinforcing steel surface has been destroyed, in the presence of water and oxygen on the surface of the reinforcement, corrosion will occur and propagate. The corrosion rate can vary considerably depending on temperature and humidity. Steel corrosion can be a very severe durability problem and can be attributed to the following factors (Bentur et al. 1997): 1 2
3
In many cases, concretes are made with high water/cement ratio with their specified strength achieved that leads to permeable structure. In practice, there is a strong need to optimize structures, which leads to minimized structural member sizes, and subsequently leads to a tendency to reduce concrete cover. The service environment can become more severe compared with those the structure was originally supposed to sustain. For example, de-icing by chloride salts is commonly used to meet the needs of the heavier winter traffic in cold climates, leading to higher chloride ingress in bridge decks and parking garages and a higher probability of corrosion of steel induced by chloride.
The signs of corrosion of reinforcing steel can be identified as rust stains and minute cracking over the concrete surface. This can be attributed to the increase in volume associated with the formation of the corrosion products and leaching of the rust. If repairs are not undertaken at the early stage when these corrosion damages occur, the corrosion of the steel will proceed further, causing severe damage through forming a longitudinal crack in parallel to the underlying reinforcement, delamination and spalling of concrete cover, as well as exposure of the steel and reduction of its cross-section that may cause a safety hazard. The damage cost of corrosion of reinforcing steel can be huge. For example, a survey by the China Academy of Engineering in 2002 reported that the annual cost due to corrosion of reinforcing steel reached 100 billion Chinese Yuan in China. So it is desirable to know clearly the mechanism of corrosion of steel and figure out the possible methods for repairing the damage caused by corrosion of reinforcing steel. There are basically two kinds of corrosion of reinforcing steel in concrete: carbonation-induced corrosion and chloride-induced corrosion. Generally speaking, chloride-induced corrosion is more serious than carbonationinduced corrosion. 2.3.4.1 Carbonation-induced corrosion Carbonation-induced corrosion is caused by the general breakdown of passivity by neutralization of the concrete. Carbonation occurs as a result of penetration of carbon dioxide from the atmosphere. In the presence of moisture, this forms carbonic acid which neutralizes the alkalinity of the cement matrix. The reduction in alkalinity destroys the passive environment and leaves the reinforcement in a condition where it is susceptible to corrosion.
Degradation of reinforced concrete structures
39
Carbonation-induced corrosion can take place on the whole surface of steel in contact with carbonated concrete. The rate of carbonation increases with an increase in the concentration of carbon dioxide especially in concretes with a high water/cement ratio, carbon dioxide and moisture creating the carbonic acid agent. The rate of carbonation also depends on both environmental factors (humidity, temperature, concentration of carbon dioxide) and factors related to the concrete (mainly its alkalinity and permeability) (Bertolini et al. 2004). The rate of carbonation varies with the humidity of the concrete. As we know, in a totally dry or wet environment, there is no carbonation. The carbonation rate may be correlated to the humidity of the environment (ibid.). The interval of relative humidity most critical for promoting carbonation is from 60– 70 percent. It has also been found that the microclimate plays an essential role on real structures. The carbonation rate may vary from one part of a structure to another or passing from the outer layers of the concrete to the inner ones, thus the carbonation of concrete can be very variable even in different parts of a single structure. As the concentration of carbon dioxide increases in the environment, the carbonation rate increases. It has been suggested that, based on experimental findings, the porosity of carbonated concrete, with a high concentration of carbon dioxide, is higher than that obtained by exposure to a natural atmosphere. If the other conditions are similar, temperature has the most important influence on carbonation rate. In common with most chemical reactions, the carbonation reaction can be accelerated by increased temperature. The quality of the concrete is regarded as the most important parameter controlling the rate of carbonation (Bentur et al. 1997). The quality of the concrete is a function of the composition of the binder (i.e. whether Portland cement or blended cement was used), the water/cement ratio, the water/binder ratio, and the curing conditions. The permeability of concrete has a remarkable influence on the diffusion of carbon dioxide and thus on the carbonation rate. The penetration of carbonation slows down as the decrease in the water/cement ratio due to the decrease of the capillary porosity of the hydrated cement paste. The type of cement also influences the carbonation rate. For blended cement, hydration of pozzolanic materials or GGBS leads to a lower carbon hydroxyl content in the hardened cement paste which may increase the carbonation rate. Consequently, the depth of carbonation is greater for the blended cements than for the Portland cement concrete when compared on the basis of equal water/binder ratios (ibid.). The fly-ash blended cement concretes show higher corrosion rates compared with Portland cement concrete with the same water/binder ratio (ibid.). Nevertheless, the water/cement ratio and curing are the most important factors influencing the rate of carbonation. The denser structure of concrete may slow down the diffusion of carbon dioxide. If cured properly, the lower alkalinity of cements with the addition of fly ash or blast furnace slag can be compensated for by the lower permeability of their cement pastes.
40 Degradation of reinforced concrete structures The chemistry of carbonation is that carbon dioxide in the air in the presence of moisture reacts with hydrated cement minerals, carbon dioxide and moisture creating the carbonic acid agent. That is to say, carbon dioxide molecules that penetrate into the concrete can react with solid calcium hydroxide, with C–S–H gel, and with the alkali and calcium ions in the pore solution. Calcium hydroxide is the hydrate in the cement paste that reacts most readily with carbon dioxide. Consequently, the Ca(OH)2 carbonates to CaCO3. The reaction can be written schematically as: CO2 + Ca(OH)2 == CaCO3 + H2O
(2.28)
This is the reaction of main interest, especially for concrete made of Portland cement, even though the carbonation of C–S–H is also possible when calcium hydroxide becomes depleted. Other cement compounds can be decomposed, namely hydrated silica alumina with ferric oxide being produced. Carbonation may cause the concrete to shrink. In the case of the concrete obtained with Portland cement, carbonation may even result in increased strength (Bertolini et al. 2004). Carbonation itself is not a problem, but it is the consequences of this on reinforced concrete which has major repercussions. Those chemical reactions during carbonation result in a drastic decrease in the alkalinity of the concrete, so that the alkalinity of the concrete is reduced from average values of 12–14 down to 8 or 9 with the consumption of calcium hydroxide. It is the high pH values which protect the steel reinforcement from corrosion. The reduction in alkalinity destroys the passive environment and leaves the reinforcement in a condition where it is susceptible to corrosion. Therefore, when carbonation penetrates through the rebar cover, and oxygen and moisture are present, corrosion will take place. The depth of carbonation in reinforced concrete is an important factor in the protection of the reinforcement; the deeper the carbonation, the greater the risk of corrosion of steel. The second consequence of carbonation is that chlorides bound in the form of calcium chloroaluminate hydrates and otherwise bound to hydrated phases may be liberated, making the pore solution even more aggressive (ibid.). The corrosion rate is usually expressed as the penetration rate of carbonation and is measured in µm per year. The depth of penetration of carbonation into concrete is proportional to the square root of time. The corrosion rate can be considered negligible if it is below 2 µm/year, low between 2 and 5 µm/year, moderate between 5 and 10 µm/year, intermediate between 10 and 50 µm/year, high between 50 and 100 µm/year and very high for values above 100 µm/year (ibid.). In high-quality concrete, the rate of carbonationinduced corrosion is negligible for relative humidity below 80 percent. It is then assumed that corrosion propagates only while concrete is wet (i.e. R.H. > 80 percent). Corrosion rate tends to decrease with time. Besides, corrosion products can reduce corrosion rate (Alonso and Andrade 1994). Page (1992) has illustrated the relationship between the corrosion rate in
Degradation of reinforced concrete structures
41
carbonated concrete and relative humidity of the environment, which indicates that the maximum corrosion rates, on the order of 100–200 µm per year can only be reached in very wet environment with relative humidity approaching 100 percent. For typical conditions of atmospheric exposure, i.e. R.H. = 70–80 percent, maximum corrosion rates are between 5 and 50 µm per year. 2.3.4.2 Chloride-induced corrosion The chloride-induced corrosion in structural concrete is primarily caused by the presence of sufficient free chloride ions in the matrix. Chloride can get into the concrete at the time of mixing, either as an admixture component or in chloride-contaminated aggregates or mix water, or penetrate the hardened concrete later on from external sources, such as sea water, slat spray, or de-icing salt placed on concrete pavements. Hence, the chloride-induced corrosion is caused by localized breakdown of passive film on the reinforcing steel. Chloride-induced corrosion is more serious for marine structures in coastal areas. In concretes of higher water/cement ratio, i.e. more porous concretes, the chloride ion penetration rate is substantially greater than that in dense concrete. The incorporation of certain mineral admixtures, such as silica fume and slag, can lower the rate of penetration to very low values in suitably dense concretes. When sufficient chlorides are present at the time of mixing, the corrosion may start at very early service stage. In the case of chloride in the atmosphere penetrating the concrete, the corrosion will not start until it reaches a certain level of accumulation of chlorides. The penetrated chloride ions diffuse through a concrete cover to the rebar surface first. Then sufficient quantities of chloride ions have to be accumulated. Next, when the concentration of chloride ion in concrete reaches a certain level (0.6–0.9 kg per cubic meter of concrete at pH value of 12.5–13.5), it dissolves the protective oxidized film, thus a localized breakdown of the passive film on the steel is formed where oxidation occurs and a galvanic cell is created. The local active area behaves as anodes, while the remaining passive areas become cathodes where reduction takes place. The effect of the separation of anode and cathode has significant consequences for the pattern of corrosion. In the concrete adjacent to the anodic areas, the concentration of positive ions increases, causing the pH to fall and allowing soluble iron-chloride complexes to form. These complexes can diffuse away from the rebar, permitting corrosion to continue. Some distance from the electrode, where the pH and the concentration of dissolved oxygen are higher, the complexes break down, iron hydroxides precipitate, and the chloride is free to migrate back to the anode and react further with the steel. The process thus becomes autocatalytic and proceeds with the deepening of corrosion pits rather than spreading corrosion laterally along the rebar. As the steel increases its state of oxidation, the volume of the corrosion products expands. The unit volume of Fe can be doubled if
42 Degradation of reinforced concrete structures FeO forms. The unit volume of the final corrosion product, Fe(OH)3·3H2O, can be as large as six and half times the original Fe. This expansion creates cracking and spalling inside concrete, and finally destroys the integrity of the structural concrete and causes a failure of buildings and infrastructures. As a result, chloride-induced corrosion is localized, with penetrating attacks of limited area surrounded by non-corroded areas. The threshold concentration of chloride ion, i.e. the critical level of chloride ion concentration in concrete at which the surface protective layer of reinforcing steel generated in high alkalinity can be broken, is a function of the pH value, i.e. the hydroxyl ion concentration. Hausman (1967) has suggested on the basis of measurements in Ca(OH)2 solutions that the threshold chloride ion concentration is about 0.6 times the hydroxyl ion concentration. Gouda (1970) has suggested another relationship in which the threshold chloride ion concentration for the pore solution of a given pH region is expressed in a logarithmic form. Both Hausman’s and Gouda’s data were derived from experiments in solution rather than in concrete, and other effects in concrete may influence the threshold value. In fact, most specifications and guidelines specify the total content of chloride in concrete in terms of percentage by weight of the original cement used. The permitted chloride contents in many specifications and recommendations are smaller than about 0.2 percent of the cement content of the concrete (Bentur et al. 1997). In the range of 0.2– 0.4 percent, there is risk of inducing corrosion, but not always. Sometimes, the critical level of chloride is specified in terms of weight of chlorides per unit volume of concrete, mostly in the range of 0.6–1.2 kg/m3. This kind of specification is needed when assessing the chlorides in existing structures, where the total chloride content of concrete can be determined experimentally, but not the chloride content in cement. The present specifications for chloride content recommended by the American Concrete Institute are given in Table 2.3. ACI and many other specifications even require that no calcium chloride-containing admixtures be used in the production of prestressed concrete. Table 2.3 Limit on chloride ions (% wt. of cement) 1. Prestressed concrete 2. Conventionally reinforced concrete in a moist environment and exposed to external sources of chloride 3. Conventionally reinforced concrete in a moist environment but not exposed to external sources of chloride 4. Above-ground building construction where the construction will stay dry. Does not include locations where the concrete will be occasionally wetted
0.06 0.10 0.15 no limita
Source: After ACI. Note: a No limit for corrosion control, if calcium chloride is used as admixture, a limit of 0.2% is generally recommended for reasons other than corrosion.
Degradation of reinforced concrete structures
43
Under the normal atmospheric environment, the chloride-induced corrosion rate can vary from several tens of µm per year (of steel) to localized values of 1 mm per year (of steel) as the relative humidity rises from 70 to 95 percent and the chloride content increases from 1 percent by mass of cement to higher values. High corrosion rates always appear on heavy chloridecontaining structures such as bridge decks, retaining walls and pillars in sea water. The corrosion rate increases when the temperature changes from a lower one to a higher one. Once the corrosion attack begins in chloridecontaminated structures, it can lead in a relatively short time to an unacceptable reduction in the cross-section of the reinforcement, even under conditions of normal atmospheric exposure. The lower limits of relative humidity near which the chloride-induced corrosion rate becomes negligible are much lower than those that make carbonation-induced corrosion negligible (Bertolini et al. 2004). The influence of temperature and humidity on the corrosion rate is shown in the electrochemical reactions at the steel/concrete interface and through their influence on ion transport between anodes and cathodes. It was thought that the concrete resistivity was strongly related to the corrosion rate at moderate or low temperature (Alonso et al. 1988; Glass et al. 1991). In a given set of conditions in terms of humidity and temperature and provided corrosion has initiated, the higher resistivity of blended cements results in a lower corrosion rate than that of Portland cement. 2.3.4.3 Corrosion mechanisms Corrosion of the reinforcing steel in concrete is an electrochemical process, consisting of both oxidation and reduction reactions, in which the metallic iron is converted to the voluminous corrosion product ferric oxide. The process is associated with the presence of anodic and cathodic areas and the potential difference between the two areas arising from lack of homogeneity in the surrounding liquid medium or even in the steel itself. The differences in potential are due to the inherent variation in structure and composition (e.g. porosity and the presence of a void under the rebar or difference in alkalinity due to carbonation) of the concrete cover, and differences in exposure conditions between adjacent parts of steel (e.g. concrete that is partly submerged in sea water and partly exposed in a tidal zone). In general, corrosion cells are formed due to: (1) contact between two dissimilar materials, such as steel rebar and aluminium conduit pipes; (2) significant variations in surface characteristics, including difference in compositions, residual strain due to local cold working, applied stress, etc.; or (3) different concentrations of alkalies, chloride, oxygen, etc. Four components must be present for corrosion to occur in a macro cell including anode, cathode, electrolyte and metallic path. The anode is the electrode at which oxidation occurs. Oxidation involves the loss of electrons and formation of metal ions. Corrosion occurs at the anode. The cathode is
44 Degradation of reinforced concrete structures the electrode where reduction occurs. Reduction is the gain of electrons in a chemical reaction. The electrolyte is a chemical mixture, usually liquid, containing ions that migrate in an electric field. The free electrons travel to the cathode, where they combine with the constituents of the electrolyte, such as water and oxygen, to form hydroxyl (OH−1) ions. The metallic path between anode and cathode is essential for electron movement between the anode and cathode. For the steel corrosion in concrete, the anode, cathode and metallic path are on the same steel. The electrolyte is the moisture in concrete surrounding the steel. The corrosion will stop if any of these components is removed. This is the basis for corrosion control. The process of corrosion is illustrated in Figure 2.9 with necessary notations of chemical reactions. At the anode site, the iron dissociates to form ferrous ions and electrons: Fe → Fe2+ + 2e−
(anodic reaction)
(2.29)
The electrons move through the metal towards the cathodic site but the ferrous ions are dissolved in the pore solution. At the cathodic site, electrons combine with oxygen and water to form hydroxyl ions: 4e− + O2 + 2H2O → 4(OH)−
(cathode reaction)
(2.30)
The hydroxyl ions move to the anode through the pore solution. At the anode, we have: Fe2+ + 2Cl− → FeCl2
(2.31)
FeCl2 + 2H2O → Fe(OH)2 (ferrous hydroxide) + 2HCl (Cl− regenerated)
(2.32)
Figure 2.9 Reinforcing steel corrosion and expansion of corrosion products: (a) corrosion process; (b) volume expansion of corrosion products.
Degradation of reinforced concrete structures
45
Fe2+ + 2(OH)− → Fe(OH)2 (ferrous hydroxide)
(2.33)
Fe + 3(OH)− → Fe(OH)3 (ferrous hydroxide)
(2.34)
4 Fe(OH)2 + 2H2O + O2 → 4 Fe(OH)3 (ferric hydroxide)
(2.35)
3+
It can be seen from the above equations that oxygen and water are needed for the initiation and propagation of the corrosion. There is no corrosion in a completely dry atmosphere, probably below a relative humidity (RH) of 40 percent (Mailvaganam 1992). It has been suggested that the optimum RH for corrosion is 70 to 80 percent. At higher relative humidity (RH greater than 80 percent) or under immersion conditions, the diffusion of oxygen is considerably reduced and the environmental conditions are more uniform along the steel. Consequently, there is little corrosion (Page and Treadway 1982). When corrosion occurs, ions need to travel through pores in surrounding concrete; oxygen needs to diffuse through the concrete cover; corrosion rate is therefore affected by electrical resistance, diffusivity, and cover thickness of the concrete. The extent of steel corrosion in concrete depends upon the conductivity of the electrolyte, the difference in potential between the anodic and cathodic areas, and the rate at which oxygen reaches the cathode. This controls the velocity of the anodic reaction. For steel in concrete, the strong polarization of the anodic zones under aqueous and highly alkaline conditions raises its potential close to that of the cathode causing the surface of the steel to be passivated by the formation of an oxide layer. The passivating film prevents further reaction so that the steel remains unaltered over long periods. Another modifying effect of concrete on corrosion is that of increased electrical resistivity, which reduces the flow of electrical currents within the concrete. This is particularly true of high-density concrete (Page and Treadway 1982; Verbeck 1975). Because serious corrosion is very dangerous to concrete structures, it is necessary to detect the degree of corrosion when deciding on maintenance. The detection methods include visual inspection, half-cell potential measurement, radiography, ultrasonic, magnetic perturbation/flux, and acoustic emission technique. Some commonly used methods will be introduced in Chapter 3. Because of the magnitude of the costs of rebar corrosion, significant efforts have been made to protect the steel in last decades. To prevent corrosion in concrete, it is essential to reduce the content of chloride to less than 0.6 kg/m3. In this case, the protection layer of rebar will remain intact and thus no corrosion will be initiated. In most instances, the corrosion control methods can be described as passive. Durability performance is obtained by proper design and control of the concrete cover. Such means are usually specified in design codes: minimum concrete cover thickness, the inherent concrete properties (in terms either of design strength or maximum water/ cement ratio) and the maximum allowable crack width permitted. Obviously, concrete quality is the most important parameter that controls the
46
Degradation of reinforced concrete structures
rate of carbonation and chloride ingress, hence the extent of reinforcing steel corrosion. Improving the quality of concrete has thus been considered the primary protection method. Next, we have to consider the four components of a corrosion cell. If we can cut out one of them, we can stop corrosion. It is obvious that the electrolyte in concrete is made up of the moisture condition and existing air. If we make denser concrete, there will be less chance for moisture and air to get in and thus reduce the possibility of forming electrolyte and prevent corrosion. Also, we can use cathodic protection method to protect reinforcing steel. For instance, using zinc as an anode can protect the steel because the corrosion occurs at the zinc anode. Other protection strategies include increased cover, epoxy-coated rebar, stainless steel, and corrosion inhibitors. The time-to-corrosion of embedded reinforcing steel can be significantly influenced by the amount of concrete cover over rebar. However, as the cover increases, the rebar becomes less effective and the potential for cracking due to tensile stress, shrinkage and thermal effects increases. Epoxy coating of reinforcing steel can enhance the durability performance by serving as a barrier preventing the access of aggressive species to the steel surface and providing electrical insulation. Epoxy-coating can be applied in various ways, either as a liquid or as a powder which is fused on the surface. ASTM A775/A775M-04a (Standard Specification for Epoxy-Coated Steel Reinforcing Bars) has addressed the basic requirements for epoxy-coated reinforcing steels by the electrostatic spray method. In general, the performance of epoxy-coated rebars in bridge decks and parking garages in chloride environments, where chloride de-icing salts are applied during winter, has been demonstrated to be excellent (Clifton et al. 1975; Satake et al. 1983). However, several notable problems with corrosion of epoxy-coated bars were reported in substructures off the Florida Keys in the USA (Clear 1992; Smith et al. 1993). The amount of damage to the epoxy coating prior to concrete casting was considered the major contributory factor to the poor performance (Smith et al. 1993). Thus, training is necessary to properly produce, handle and construct the coating, and repair of field damage to epoxy-coated bars. Besides, the bond properties between concrete and the epoxy-coated rebars are not as strong as that between concrete and conventional rebars, which should be improved. There are a few studies on the use of stainless steel bars. It has been shown that the chloride threshold value for initiation of corrosion in non-welded AISI 304 (a kind of stainless steel) rebar is three to five times higher than that of conventional rebar. However, welding of the bar reduced the critical chloride level by 50 percent. In addition, use of stainless rebars is an expensive solution. Zinc coating of steel (galvanized steel) is also considered a good method of corrosion resistance. It acts both as a sacrificial and barrier-type coating. But the disadvantage is that, like other metal coatings, zinc coating corrodes over time. The rate of corrosion under the given environmental conditions will determine the loss of coating thickness and the time period during which it will be effective. Generally,
Degradation of reinforced concrete structures
47
there is a fairly linear relationship between the metal thickness and the duration of its effective service life for galvanized steel exposed to industrial atmosphere (Chandler and Bayliss 1985). The stability of zinc is dependent on the pH of the surrounding solution where the zinc coating is exposed. It has been found that zinc is stable at pH below about 12.5, but it tends to dissolve at an increasing rate as the pH increases above this level. The corrosion products of zinc may be deposited at the surface of the zinc coating and seal it, thus arresting the evolution of H2 gas and leading to passivation of the zinc coating. However, if the galvanized rebars are mixed used with ungalvanized bars, the galvanized bars will be accelerated depleted. So if the galvanized and ungalvanized bars are being used in the same structure, special care should be taken to ensure complete electrical isolation of these two (Broomfield 1997). Corrosion inhibitors are considered useful not only as preventative measure for new structures but also as preventative and restorative surfaceapplied admixture for existing structures. Various corrosion inhibitors (see Trabanelli 1986) can be classified into: (1) adsorption inhibitors, which act specifically on the anodic or on the cathodic partial reaction of the corrosion process or on both reactions; (2) film-forming inhibitors, which block the surface more or less completely; and (3) passivators, which favour the passivation reaction of the steel. The mechanistic action of corrosion inhibitors is thus not against uniform corrosion but localized or pitting corrosion of a passive metal due to the presence of chloride ions or a drop in pH value (Bertolini et al. 2004). Corrosion inhibitors admixed to the free concrete can act in two different ways: these inhibitors can extend the corrosion initiation time and/or reduce the corrosion rate after depassivation has occurred (Hartt and Rosenberg 1989). Mixed-in inhibitors are regarded as more reliable since it is easier and more secure to add the inhibitors to the mix. Some laboratory testing has shown that certain corrosion inhibitors do not significantly affect the amount of chloride ion required to initiate corrosion, but can reduce corrosion rate. The field performance of these products lasts only a relatively short period and cannot be conclusive in determining their effectiveness. For example, the field tests with proprietary vapor-phase inhibitors in a parking garage with chloride-contaminated precast slabs did not show encouraging results (Broomfield 1997). Corrosion-rate measurements showed a reduction of 60 percent in areas with initially intense corrosion but also an increase of corrosion rate in areas with low corrosion rates. 2.3.5 Degradation caused by alkali–silica reaction Alkali–Aggregate Reaction (AAR) is a reaction between alkalis from cement and certain forms of silica presented in aggregate, which results in excessive expansion of concrete sections and leads to severe cracking thereafter. Two general types of attacks can occur: (1) alkali-carbonate attack with
48 Degradation of reinforced concrete structures dolomitic limestone aggregate (some argillaceous dolomites); and (2) alkalisilica attack with siliceous aggregates containing certain forms of amorphous or poorly crystalline silica (such as some chert, flint, opal, tridymite, cristoballite chalcedony, volcanic glasses and some limestones). The alkali content of cement depends on the materials from which it is manufactured and also to a certain extent on the details of the manufacturing process, but it is usually in the range of 0.4–1.6 percent. In concrete mixes there may be a contribution to alkali from other cementitious materials, such as pulverized fuel ash or ground granulated blast-furnace slag, which is present. It is well known that Na2O (sodium oxide) and K2O (potassium oxide) are present in the cement clinker in a small amount. It is thus conventional to express the results of chemical analysis of cement in terms of the oxides, Na2O and K2O. Furthermore, the alkali content in cement is generally expressed as an equivalent percentage of Na2O by mass of cement. Since the molecular weights of Na2O and K2O are respectively 62 and 94, the equivalent percentage of Na2O is calculated by the formula: %Na2Oeq = %Na2O + 0.658·%K2O
(2.36)
In the cement paste, Na2O and K2O form hydroxides and raise the pH level from 12.5 to 13.5. The concentration of these hydroxides increases as Na2Oeq increases. In such highly alkaline solutions, under certain conditions, the silica can react with alkaline to form a silicate gel, which is hygroscopic. When the gel absorbs moisture, it will swell and apply pressure to the surrounding paste, causing the concrete to crack, sometimes severely, with accompanying loss of structural integrity. The degree of AAR is affected by: (1) presence of water, if there is no water, there is no expansion; (2) alkali content, if the alkali content (Na2O and K2O) is less than 0.6 percent, there is no reaction and the concrete contains more than 3 kg/m3 of alkali can be considered to have a high alkali content; (3) concrete porosity, the internal stress may be relieved in concrete with high porosity. ASR can occur only in a moist environment: it has in fact been observed that in environments with a relative humidity below 80–90 percent, alkali and reactive aggregate can co-exist without causing any damage. With low effective alkali content in the concrete, i.e. the equivalent content of Na2O in concrete is less than 3 kg/m3, deleterious ASR can be prevented. The expansive effect, hence ASR effect, can be negligible. The extent of reaction depends on the amount of reactive silica present in the aggregate mix while the reactivity of silica minerals depends on their crystal structure and composition. The porosity, permeability and specific surface of aggregates and the presence of Fe- and Al-rich coatings can influence the kinetics of the alkali silica reaction. It should be noted that blended cements containing pozzolana, fly ash or blast furnace slag give a resulting alkaline solution of slightly lower pH. Addition of silica fume leads to the lowest pH. Hence, use of pozzolana materials such as fly ash, GGBS can even prevent damage caused by ASR. It
Degradation of reinforced concrete structures
49
has also been found that by adding calcium nitrite into the concrete mix, the concrete’s resistance to ASR can be significantly improved. However, the mechanism is not clear (Li et al. 1999; 2000). These mineral additions reduce the concentration of OH−1 ions in the pore solution of cement paste. This is because hydroxyl ions are consumed by the pozzolanic reaction occurring during hydration. Furthermore, the alkali transport is slowed down because of the lower permeability of pozzolanic and blast furnace cement paste, which helps reduce the ASR. Finally, the hydration products of the mineral additions bind alkali ions to a certain extent, preventing them from taking part in the reaction with the silica. The temperature also influences the alkali silica reaction. Normally the ASR increases as the temperature increases. AAR can be deleterious to concrete due to the expansion and possible cracking of the concrete associated with the reaction. What’s more, development of alkali silica reaction may be very slow and its effects may show even after long periods (up to several decades). Consequently, cracking caused by alkali silica reaction usually takes many years and is often preceded by pop-outs on the concrete surface. It is believed that AAR was first identified in 1940 in California, by Stanton (Mehta and Monteiro 2006), but only limited numbers of examples were observed in practice until fairly recently. AAR causes deleterious expansion and cracking of the concrete and reduces the tensile strength, which may have consequences for the structural capacity. The significance of AAR or ASR for concrete structures have led to a surge in research activities: (1) to determine the exact nature of the reaction (which is not fully understood as yet); (2) to define the acceptable limits of alkali content, moisture content, and reactive aggregate content; and (3) to determine methods to reduce the degree of destructive expansion. Extensive research has been made in two directions. One is the development of testing methods, which are designed to reveal whether an aggregate is potentially reactive and can cause abnormal expansion, and cracking of concrete. The other is the development of effective methods to prevent the damage induced by AAR. Although it is possible to determine what types of aggregates have a trend of AAR, it is impossible to predict whether their use will result in excessive expansion or not. It has been found that a critical amount exists for each type of aggregate, and only this can result in serious expansion, an amount smaller or larger than this value will not cause significant swelling. Measuring the expansion of test specimens has been considered the most dependable way to evaluate aggregate reactivity, and a number of test procedures have been devised. The testing methods commonly used include standard test methods such as the mortar bar test (ASTM C227), the rock cylinder method (ASTM C586), and rapid test methods such as ASTM C289. One of the disadvantages of the standard testing methods is that most of them are very time-consuming, which is obviously incompatible with the demands of the construction industry. On the other hand, the rapid testing method,
50
Degradation of reinforced concrete structures
ASTM C289, only gives the potential reactivity of the aggregate, but it does not necessarily predict expansion in a real situation. To overcome the limitations of the above-mentioned test methods, other rapid testing methods have been developed, such as the dynamic modulus test and the gel fluorescence test. Dynamic modulus can be obtained by measuring resonant frequency and pulse velocity. It has been shown that it can provide a good indication of deterioration due to AAR. The measurement can even detect deterioration before any expansion and visible cracking occurs. It is also sensitive to the changes in environmental conditions, which activate or suppress the AAR. In the gel fluorescence test method, 5 percent solution of uranyl acetate is applied to the surface of the specimen, then the specimen is viewed under an ultraviolet (UV) light. A yellowish green fluorescent glow means that AAR is present. Since AAR can cause significant deterioration and damage to concrete structures, many research studies have been conducted to decrease the effect of the reaction. There are numerous recommendations for minimizing the risk caused by AAR. These recommendations include: 1 2
3
4
Use non-deleterious aggregate and/or non-reactive aggregate when the alkali content of the cement is high. Use low-alkali Portland cement or blended cements with sufficient amounts of fly ash or slag when the silica content of aggregate is high (more than 3.0 kg/m3). Keep the concrete dry (relative humidity of the concrete less than 80 percent). However, the choice of types of cements and aggregates on a construction site is usually very limited, and the environment surrounding the concrete is obviously unchangeable. In many cases, the contents of silica in aggregates and/or alkalis in cement paste cannot be reduced effectively, either. The only effective way to reduce the risk of AAR is to control moisture migration in the concrete since no AAR will occur in dry concrete even there is some silica present. Controlling moisture diffusion in concrete can be implemented in two different ways. One is to control the local moisture diffusion around the boundary of each aggregate, that is, to control the moisture exchange between the aggregate and surrounding cement paste. The other is to control the diffusion of moisture into and out of the surface of concrete members. Local diffusion control coating. The control of local moisture diffusion is very important because AAR occurs right on the boundary of the aggregate. Distributions of the chemicals, Na2O, SiO2, K2O, and CaO, around the aggregate boundary have shown quantitatively that the reaction rim is in the range of 300 micrometers. Local diffusion control coating is developed based on the crystallization technique. The coating product consists of powders of finely ground rapid setting Portland cement, treated silica sand (de-alkaline silicate), and proprietary chemical additives. They are mixed with water to form slurry. The slurry will
Degradation of reinforced concrete structures
5
51
be applied to the reactive aggregate before the cast of concrete specimens. A coating is then formed around the aggregates. Application of this technique is not aimed at completely eliminating AAR but at reducing and slowing down the rate of AAR. As a result of a slow reaction, the product of the reaction can be accommodated and deposited in large capillary pores. Thus the detrimental damage due to the expansion will be avoided. Global diffusion control coating. This method applies a sunlight curing hydrophobic weather-resistant coating. This is a newly developed coating, which can be applied to the surface of concrete members to reduce moisture diffusion. In addition to the high resistance to moisture penetration, the coating has other two useful features. One is its ability to reduce or to eliminate VOC (Volatile Organic Compounds) which are detrimental to environment. The other is that it can be cured under low power UV. UV curing is very important for long-term applications since it yields paint or coatings with excellent durability. However, conventional UV curing requires a special high power UV lamp, while this coating requires only natural light from the sun, that is, sunlight curing.
The alkali–silica reaction is most widespread and best understood. In the case of alkali–silica reaction, alkali hydroxides in the hardened cement paste liquor attach to the silica to form an unlimited swelling gel, which draws in any free water from osmosis and thus expands, disrupting the concrete matrix. Expanding gel products exert internal stress within the concrete, causing characteristic map cracking of unrestrained surfaces, but cracks may be directionally oriented under conditions of restraint imposed by reinforcement, pre-stressing or loading. Cracking resulting from alkali–silica reactions was long held not to pose any structural threat to affected members. Consideration must always be paid to the effects of deep penetrating cracks on the durability of reinforcement and to the self-stress induced by the expansive reactions caused by alkali reactivity. This may be advantageous in confined sections of normally reinforced concrete members, but could prove catastrophic in the case of pre-stressed structures. Avoidance of the problem caused by alkali reactivity prior to construction is of greatest importance. This could be achieved by avoiding the use of reactive aggregates, by the use of low alkali Portland cement, of slag cement or of pozzolanic admixtures. As the swelling phenomenon is dependent on water absorption, most control measures are taken by decreasing water ingress to the concrete by the use of surface coatings or impregnation materials (Campbell-Allen and Roper 1991). 2.3.6 Degradation caused by sulfate attack Sulfate attack is one of main factors causing deterioration of concrete durability. In most cases, the sulfates are external to the concrete, i.e. in a
52 Degradation of reinforced concrete structures solution in ground water or trade effluent and from contaminated aggregates (e.g. in the Middle East). Portland cement itself also contains sulfate as gypsum. If saline water, such as sea water, is used for mixing concrete, it definitely brings sulfate to the mixture. Solutions of sulfates of sodium, potassium, magnesium, and calcium are commonly slats that may bring severe deterioration to concrete. The total sulfate content in concrete mixture becomes a controlling factor of the degree of sulfate contact. The acceptable concentration of sulfate in concrete is about 4 percent by weight of cement. Sulfate attack of concrete is generally regarded as an expansion that can cause concrete cracking due to the reaction of sulfate with some hydration products in cement paste. When concrete cracks, its permeability increases and the aggressive water penetrates more easily into the interior, thus accelerating the process of deterioration. Hence, the deterioration of concrete due to the sulfate attack can be considered a complicated phenomenon of physical and chemical process. Sulfate attack of hydrated cement takes place by the reaction of sulfate ions with calcium hydroxide and hydrated calcium aluminates to form gypsum and ettringite and the formation of these two products is responsible for expansion and cracking of concrete. Sulfate attack can also manifest itself as a progressive loss of strength of the cement paste due to loss of cohesion between the hydration products and loss of adhesion between the hydration products and the aggregate particles in concrete. The sulfate reacts with the tricalcium aluminate (C3A) in the cement to form the compound ettringite, which causes the expansion of concrete. Calleja (1980) has pointed out that in all the sulfates, magnesium sulfate attack has the most damaging effect because Mg2+ and Ca2+ ions associate well, they have equal valence and similar ionic radii, which can lead to a reaction between magnesium sulfate and C–S–H gel. The deterioration of Portland cement concretes exposed to sulfates normally may be ascribed to the following reactions (CampbellAllen and Roper 1991): (i)
The conversion of calcium hydroxide derived from cement hydration reactions to calcium sulfate, and the crystallization of this compound with resulting expansion and disruption. Ca(OH)2 + SO42− + 2H2O → CaSO4·2H2O + 2OH−
(2.37)
(ii) The conversion of hydrated calcium aluminates and ferrites to calciumsulfo-aluminates and sulfo-ferrites or the sulfate enrichment of the latter minerals. The products of these reactions occupy a greater volume than the original hydrates, and their formation tends to result in expansion and disruption. C3A + 3CaSO4 + 32H2O → 3CaO·Al2O3·3CaSO4·32H2O (iii) The decomposition of hydrated calcium silicates.
(2.38)
Degradation of reinforced concrete structures
53
In the presence of calcium sulfate, only reaction (ii), can occur, but with sodium sulfate, both (i) and (ii) may proceed. With magnesium sulfate all three may occur. The severity of attack depends on the type of sulfate. Calcium sulfate undergoes expansion reaction with ettringite (arising from the calcium aluminate in cement), which gives rise to greater expansive effects than gypsum. Part of the formed ettringite is commonly located in the interface between paste and aggregate, resulting in loss of bond. Sodium sulfate (Na2SO4) also reacts with calcium hydroxide to form gypsum, which reduces paste strength and stiffness. Magnesium sulfate (MgSO4) reacts to form gypsum and destabilizes C–S–H, the strength-governing phase in the cement paste. Severity of attack therefore increases from calcium sulfate to sodium sulfate to magnesium sulfate. As a reference, ACI 201.2R-92 (1994a) has classified the severity of exposure, as given in Table 2.4. Sulfates in solution can combine with the tri-calcium aluminate (C3A) in Portland cement, to form a sulfoaluminate hydrate (ettringite) in the form of needle-like crystals, causing expansion of the matrix which turns white and become soft. The ettringite occupies a greater volume within the concrete than the calcium aluminate hydrates. The expansion generates tensile stresses in the cement paste and as a result cracks develop in the concrete. Cementitious repairs in these conditions should use sulfate-resisting cement, which has a low C3A content. Attack on dense concrete will be a surface effect so rich, well-compacted mixes must be used. The form of sulfate present in uncontaminated groundwater is normally calcium sulfate, which has limited solubility. Some other salts, such as magnesium sulfate, are much more readily soluble in water and can form stronger solutions, so they are more dangerous (Allen et al. 1993). For Portland cement-based concrete, the resistance to sulfate attack depends primarily on the type and amount of cement in the concrete, the type and amount of mineral additives in the concrete, and the permeability and diffusivity of the concrete. Low permeability and diffusivity provide the best defense against sulfate attack by reducing sulfate penetration. To achieve a low permeability and diffusivity, the water/powder ratio should be as small as possible and pozzolanic additives that can reduce the calcium hydroxide content and refine the pore structure of the matrix through Table 2.4 Classification of severity of sulfate environment according to ACI 201.2R92 Exposure
Mild Moderate Severe Very severe
Concentration of water-soluble sulfate expressed as SO4 In soil (%)
In water (ppm)
2.0
10000
54 Degradation of reinforced concrete structures pozzolanic reaction should be used. Reducing the water/cement ratio can make concrete less permeable and more difficult for sulfates to penetrate. It has been proved to be more effective than reducing the calcium aluminate content for improving the sulfate resistance of concrete. The severity of sulfate attack depends on the content of C3A and, to a lesser extent, of C4AF in the cement. Sulfate-resisting cement has a low tricalcium aluminate content and hence less potential for expansive reaction and better resistance to sulfate attack. Reducing the amount of calcium hydroxide and hydrated calcium aluminate is also helpful in improving the sulfate resistance of concrete. Blended cements with pozzolanic materials or blast furnace slag show enhanced resistance to sulfate attack (Mehta 1988a). For example, lowcalcium fly ash is an effective blending material to combat the sulfate attack of concrete. Also the incorporation of silica fume into a concrete mixture greatly improves the sulfate resistance, owing to the reduced amount of gypsum formation, compared with mixtures that do not contain silica fume. Also, the incorporation of silica fume can largely reduce the permeability of concrete, thus improving the sulfate resistance. The mixture of silica fume and fly ash should be a much better choice for production of high performance concrete. On the other hand, Mehta (1988a) has noted that it is the mineralogy rather than the chemical composition which determines the resistance to attack, and empirical guidelines based on chemical compositions of a mineral additive do not prove to be reliable. Nevertheless Mehta (1980, 1988b) has concluded that as a first approximation, a blended cement will be sulfate resistant when made with a highly siliceous natural pozzolanic or low-alumina fly ash or slag, provided that the proportion of the blending material is such that most of the free calcium hydroxide can be used up during the course of cement hydration. Under severe conditions, hydraulic cements other than Portland cement should be used, for example supersulfated cement and high-alumina cement. Supersulfated cement offers very high resistance to sulfates, especially if its Portland cement component is of the sulfate-resisting variety. It should be noted here that high-alumina cement should not be used in continuously warm, damp conditions, or in mass construction from which the heat of hydration cannot be easily dissipated. High-pressure steam curing improves the resistance of concrete to sulfate attack due to the change of C3AH6 into a less reactive phase, and also to the removal of Ca(OH)2 by the reaction with silica. The concrete attacked by sulfate shows a whitish appearance. Usually, damage starts at the edges/corners, followed by progressive cracking/ spalling. Rate of sulfate attack also increases with concentration of sulfate and replenishment rate (e.g. sulfate attack to concrete is faster in flowing groundwater due to faster replenishment rate). Tests on sulfate resistance are normally conducted by storing specimens in a solution of sodium or magnesium sulfate, or a mixture of the two. The test may be accelerated with wetting/drying cycles that will induce salt crystallization in the pores. Effect of exposure can be estimated from: (1) change in dimension; (2) loss of
Degradation of reinforced concrete structures
55
strength; (3) change in dynamic modulus of elasticity; and (4) loss of weight. The test method of ASTM C1012-04 uses the immersion of well-hydrated mortar in a sulfate solution, and considers excessive expansion as a criterion failure under sulfate attack. But this method is only for mortar not concrete. Besides, this method is slow and normally takes several months. As an alternative, ASTM C 452-89 prescribes a method in which a certain amount of gypsum is included in the original mortar mix. This speeds up the reaction with C3A but this method is not appropriate for use with blended cements. The criterion of sulfate resistance in this method is the expansion at the age of 14 days. 2.3.7 Acid attack Acid solutions, both mineral and organic, are about the most aggressive agents for concrete since Portland cement concrete is highly alkaline, Depending on the type of acid, the attack can be mainly an acid attack, or a combination of acid followed by a salt attack. In normal concrete, the hydrated components in the cement matrix of concrete are in equilibrium with the pore solution having a high pH value, due to the presence of OH−1. When concrete comes into contact with acid solutions, these compounds may dissolve at a rate depending on the permeability of the concrete, the concentration and the type of acid. Moreover, acids can also attack calcareous aggregates such as limestone in concrete. If the acid penetrates the concrete through cracks down to the steel bars, the steel bars can corrode and spall the concrete, causing deterioration of concrete. Owing to the highly basic character of Portland cement, an acid cannot penetrate dense concrete without being neutralized as it travels inwards. Therefore, it cannot cause deterioration in the interior of the specimen without the cement paste on the outer portion being completely destroyed. The rate of penetration is thus inversely proportional to the quantity of acid-neutralizing material, such as the calcium hydroxide, C–S–H gel, and limestone aggregates (Mailvaganam 1992). In summary, concrete can be attacked by liquids with a pH value below 6.5 but the attack is severe only at a pH value below 5.5. When a pH value is below 4.5, the attack is very severe. Acid can naturally occur from rain in industrial regions, which is dilute solutions of CO2, and SO2, and CO2 bearing groundwater. Water containing dissolved carbon dioxide can be a common acid source in concrete. With the presence of CO2 in water, calcium hydroxide can be converted into: (1) calcium carbonate, which is stable but it will reduce alkalinity of concrete; (2) calcium bicarbonate, which can dissolve in water and leach out of the concrete. Once reaction occurs on the surface, CO2 needs to travel a longer distance for further reaction. Similar to dry oxidation, its penetration rate is approximately proportional to the square root of time. So the acidic attack progresses at a rate approximately proportional to the square root of time. If these acids come into contact with concrete they can react with the alkali
56 Degradation of reinforced concrete structures present in the cement hydrates to produce soluble products that can be removed by solution. In flowing water, the reaction products are carried away, exposing fresh surfaces to attack. Under static conditions, the water adjacent to the structure may become saturated. If this happens, the reaction ceases after only surface attack. The degree of acidic attack on the cement matrix depends on the solubility of the salts formed and thus on the nature of the anions involved (Bertolini 2004). For more soluble reaction products, the attack rates are higher than for insoluble products. For hydrochloric acid, soluble calcium chloride is formed, while with sulfuric acid, calcium sulfate (gypsum) is formed, which is much less soluble. Sulfuric acid with the presence of SO2 is particularly aggressive because the sulfuric acid attack Ca(OH)2 and C–S–H takes place apart from the attack in the alumina phase. Though various physical and chemical tests on the resistance of concrete to acids have been developed, currently there are no standard procedures. In general, acid attack can be reduced by: (1) use of low water/cement ratio to reduce permeability; (2) use of low cement content to reduce C–S–H; and (3) use of pozzolanic materials, such as ground granulated blast-furnace slag, pozzolanas, and especially silica fume, to reduce the Ca(OH)2 content. Where the acid attack possibility is very high, such as floors of chemical plants, the surface can be treated by sodium silicate (water glass) which reacts with Ca(OH)2 and forms calcium silicate to block pores, so that reduces the acid attack. 2.3.8 Frost attack Concrete is a porous material. As the cement and water in fresh concrete react to form a hardened paste binding the coarse and the fine aggregates together, voids are left in the originally water-filled space among the cement grains. These voids are known as capillary pores with a size range from approximately 5 nm to 1 µm and sometimes even larger. In addition to capillary pores, cement paste also contains a significant volume of smaller pores called gel pores. Water contents and capillary forces in small volumes of these pores are very important to the durability of hardened concrete, especially for those subjected to repeated cycles of freezing and thawing, which can cause disintegration of concrete surface layers. Concrete deterioration caused by freezing and thawing is linked to the presence of water in concrete but cannot be explained simply by the expansion of water on freezing. While pure water in the open freezes at 0 °C, in concrete the water is really a solution of various salts so that its freezing point is lower. Moreover, the temperature at which water freezes is a function of the size of the pores (Figure 2.10). In other words, the temperature at which water can freeze in concrete pore decreases as the size of the pore decreases. In concrete, pore sizes have a wide range so that there is no single freezing point. As an example, for saturated Portland cement paste, free water in pores larger than 0.1 mm freezes between 0 and −10 °C; water in
Degradation of reinforced concrete structures
57
Figure 2.10 Relationship between the size of capillary pores and the temperatures at which ice formation (from pure water) is possible inside their pores.
pores between 0.1 and 0.01 mm freezes between −20 and −30 °C and gelwater (pores less than 10 nm) freezes below −35 °C (Beddoe and Setzer 1988; 1990). Freezing begins in the outer layers and in the largest pores and extends to the inner parts and to smaller pores only if the temperature drops further. Specifically, the gel pores are too small to permit the formation of ice, and most of the freezing takes place in the capillary pores. We can also note that larger voids, arising from incomplete compaction, are usually airfilled and are not appreciably subjected to the initial action of frost. During the freeze–thaw cycle attack on concrete, the presence of de-icing salts, like calcium and sodium chloride, in contact with concrete is a detrimental factor. The outer layers where these salts are present are more strongly affected by frost despite their effect in lowering the freezing point, probably due to increased water saturation caused by their hygroscopic effect, with the early appearance of scaling and detachment of cement paste which covers the aggregate. When water freezes, there is an increase in volume of approximately 9 percent. Pockets in concrete that can fill with water, or with materials that absorb water are also source of trouble because the water in them will expand if it freezes, disrupting the surrounding concrete (Allen et al. 1993). As the temperature of concrete drops, freezing occurs gradually inward and puts hydraulic pressure on the unfrozen water in the capillary pores due to the volume expansion of ice. Such pressure, if not relieved, can result in internal tensile stresses that may cause local failure of the concrete. On subsequent thawing, the expansion caused by ice is maintained so that there
58 Degradation of reinforced concrete structures is now new space available for additional water which may be subsequently imbibed. During re-freezing, further expansion occurs. Thus repeated cycles of freezing and thawing have a cumulative effect. It is the repeated freezing and thawing, rather than a single occurrence of frost that causes damage. Frost action is an important factor causing concrete degradation in cold region. When the solutions contain de-icing chemicals, though the de-icing salts can lower the temperature of ice formation, which may be viewed as a positive effect, they may also bring the following negative effects: (1) an increase in the degree of saturation of concrete due to the hygroscopic character of the salts; (2) an increase in the disruptive effect; (3) the development of differential stresses as a result of layer-by-layer freezing of concrete due to salt concentration gradients; (4) temperature shocks; and (5) salt crystallization in supersaturated solutions in pores (Mehta and Monteiro 2006). Overall, the negative effects far outweigh the positive effect. There are two other processes that can increase hydraulic pressure of the unfrozen water in the capillaries. First, since there is a thermodynamic imbalance between the gel water and the ice, diffusion of gel water into capillaries can lead to a growth in the ice body and thus to an increase of hydraulic pressure. Second, the hydraulic pressure is increased by the pressure of osmosis caused by local increases in solute concentration due to the removal of frozen (pure) water from the original solution. The extent of damage caused by repeated cycles of freezing and thawing varies from surface scaling to complete disintegration as layers of ice are formed, starting at the exposed surface of the concrete and progressing through its depth. In general, concrete members that remain wet for long periods are more vulnerable to frost than any other concrete. It is clear that the hydraulicpressure mechanism of frost damage has more severe consequences in a system of fully saturated pores, because in that case the pressure can only be released if the microstructure expands, which may quickly result in cracking. In general, the loss of mass or the decrease of dynamic modulus is used as the index of degradation. The general influence of saturation of concrete, in the deterioration mechanism, is related to a value known as the critical saturation, below which the concrete is quite resistant. The critical saturation depends on the size of the body, on its homogeneity, and on the rate of freezing. Frost resistance is determined by the number of freeze–thaw cycles that a particular concrete can withstand before reaching a given level of degradation (Bertolini 2004). Another important parameter to frost resistance is the water/cement ratio, which largely determines the porosity of the cement matrix. In order to prevent a concrete from damage caused by repeated cycles of freezing and thawing, air-entraining agent can be used. An air-entraining agent is a chemical admixture that can deliberately entrain the air into cement paste with a closed space ( Eold, σnew > σold
(4.2)
Figure 4.2 Stress distribution in two materials with different moduli of elasticity.
160
Conventional repair and strengthening
This means that the stress distribution will not be uniform in the repaired section. Before the repair, the stress in the column was uniformly distributed. After the repair, the load can be considered as being transferred from the old concrete to the new concrete. If the strength of the new material with higher modulus is not sufficient to bear the load, the new materials may fail earlier than the old concrete. This is one of the important mechanisms that are responsible for pre-matured failures of repair work in the new material. On the other hand, if a repair material (e.g. a polymer mortar) with low modulus of elasticity but high compressive strength is used to repair an old concrete, it could lead to the failure of the old concrete due to the stress transfer to the old concrete, assuming that the old concrete has higher modulus of elasticity than the new concrete. The failure mechanism can be explained by the same model shown in Figure 4.2. In this case, the calculation of the stresses remains the same as Eq. (4.1), and since Enew < Eold, we have σnew < σold, which means more stress would be transferred to the old concrete. So, the load carried by the old concrete would be even higher than before. Another possibility of repair failure due to the non-uniform stress distribution is the bond failure between the new and old material. It should be noted that, except for external loads, shrinkage or thermal expansion and contraction can also cause loss of bond between the repair material and old concrete when the moduli of elasticity of the two materials are significantly different. Table 4.4 shows moduli of elasticity of various repair materials (Emmons 1994). It will be wise to select a repair material with the same or similar modulus of elasticity compared with the old concrete. Table 4.4 Moduli of elasticity of repair materials (Emmons 1994) Materials
psi (×106) Mpa (×104) Materials
Portland Cement Mortar
3.4 2.3 Preplaced-Aggregate Concrete 3.8 2.6 Portland Cement Concrete
3.8 2.6
Methyl-methacrylate Concrete 3.0 2.0 Epoxy Mortar 2.2 1.5
Magnesium Phosphate Cement Concrete Microsilica Modified Portland Cement Concrete Polymer Modified Portland Cement Mortar with Non-sag Filler Latex Modified Portland Cement Concrete Shotcrete
psi (×106) Mpa (×104) 3.2 2.2 4.0 2.8 2.5 1.7 2.5 1.7 3.8 2.6
Conventional repair and strengthening
161
4.2.1.2 Compatibility in chemical stability If the repair material and the existing material have different chemical properties, the repaired concrete may not be durable. For instance, if the existing concrete is made of low alkaline cement and reactive aggregate, and the repair material is made of high alkaline cement, the so-called alkali–silica reaction (ASR) between the new cement and old aggregate may take place. ASR will be discussed in detail in a later section. Another example is that shrinkage-compensating mortar or concrete is used as repair material while existing concrete is normal concrete. In this case, special care should be taken to ensure the amount of expansion of the repair material is properly selected and controlled. Otherwise, the expansion may cause cracking in the old concrete. This is especially important for underwater repair work. If the expansion of the shrinkage-compensating cement is not well controlled, cracks may appear in the repair system, resulting in leakage of the concrete structures. Careful consideration on the chemical contents of the repair material must be made when the repair job deals with reinforcing steel or other embedded metals. Usually the chloride content in the old concrete is relatively high for the concrete exposed to de-icers and the chloride content in the new concrete is very low. A large chloride concentration gradient exists between the new and old concrete, which may form a “concentration cell” for the onset of steel corrosion in concrete. This “concentration cell” will accelerate the rate of corrosion and cause premature failure of the patch or adjoining concrete. The chloride concentration gradient is a driving force in the accelerated diffusion of the chloride in the old concrete into the new concrete, which causes the early onset of steel corrosion in the new concrete. For these cases, there are two types of methods that can be used to prevent the premature failure of the repaired system. One is to lower the chloride in the old concrete by using available chloride removal techniques (which are usually expensive). The other is to install a cathodic protection system in the new concrete. These methods will be discussed in detail in Section 4.3.6.
4.2.1.3 Compatibility in transport properties Transport properties of the new and old materials must also be compatible. There are usually two transport properties of concrete that are worthy of consideration: permeability and diffusivity. Permeability refers to the ease that fluids, both liquids and gases, can enter into and move around in the concrete. According to Darcy’s law, the flux of a fluid is proportional to the pressure gradient in the fluid, and the proportionality constant is the permeability. Diffusivity refers to the ease that species (such as chloride ions) penetrate into concrete. According to Fick’s law, the flux of a diffusing species is proportional to the concentration gradient of the species, and the proportionality constant is the diffusivity.
162
Conventional repair and strengthening
The permeability and diffusivity of a repair material should be as close to that of exiting concrete as possible. If there is a large difference in the transport properties, some long-term durability problems may occur. For instance, when the repaired structure undergoes wet-dry cycling (e.g. at the tidal zone), the differences in the permeability and diffusivity of the new and old concrete result in concentration gradients of the oxygen and chloride in the repaired concrete, which may result in the corrosion of reinforcing steel in the concretes. The transport properties of concrete depend on its internal microstructure, which in turn depends on the concrete mix design. Usually the repair material has lower permeability and diffusivity than that of the existing concrete. 4.2.1.4 Other considerations regarding compatibility In addition to the compatibility considerations discussed above, there is another compatibility related to the electrical impedance difference between the repair material (such as macromolecule compound mortar) and the host material (such as normal concrete). In the research community, there is still some disagreement over whether a large difference in impedance may reduce or accelerate the steel corrosion in concrete. In other words, it is not clear if a large impedance difference or a small difference is better for enhancing the corrosion resistance of the repaired structure. In general, there are similar electrical impedances between the repair material such as the Portland cement-based materials, and the host material such as normal concrete. It is also worthwhile pointing out that the anticipated service conditions of the repaired system and the weather conditions when the repair material is applied play an important role in the final choice of the right repair material. The following factors relating to application and service conditions should be considered in choosing the repair material (Mailvaganam 1992): 1 2
3
4
The moisture content in the substrate. This will be used to determine the mix design of the repair material. The temperature at the time of application of the repair; for instance, the hydration and curing rate of cementitious materials may be influenced by the environment temperature. For the same repair work, winter construction and summer construction make a significant difference in terms of selection of materials. The maximum and minimum service temperature of the structure, the range of service temperature will induce possible thermal movements and build up stresses. The location of the repair section should be considered; for instance, vertical or horizontal surface for repair should be considered for the mix design of the repair material.
Conventional repair and strengthening 5
6
7 8
9
163
Turn-around time; for instance, if the repair is subjected to early loading, the material with high early strength may be used (e.g. type I vs. type III cement). Chemical exposure; for instance, if there are acids, alkalis, sulfates and so on in the environment, special cement, polymers, or additives should be considered. Exposure to traffic, a repair material with good abrasion resistance may be needed for overlay exposed to heavy traffic. Importance of appearance; for instance, for repairing a stamped concrete slab, the color matching between the new material and old concrete is important. Life of repair, for a temporary repair work, some of the compatibility requirements may be relaxed.
4.2.2 Standard testing methods and requirements for repair materials Depending on the type of repair material selected for a repair project, several specific testing methods can be used to evaluate the properties of the material. For cementitious repair materials, ASTM C928 (Standard Specification for Packaged, Dry, Rapid-hardening Cementitious Materials for Concrete Repairs) is the current industrial standard in the US for the evaluation of the performance of cementitious repair materials. The required tests in ASTM C928 include: • • • • •
compressive strength: ASTM C 109/C39 slump of hydraulic cement concrete: ASTM C 143 length change of hardened hydraulic-cement mortar and concrete: ASTM C 157 scaling resistance of concrete: ASTM C 672 bond strength: ASTM C 882.
The performance requirements of repair materials in ASTM C928 are listed in Table 4.5. One can see that the requirements include most of the compatibility conditions discussed earlier in the chapter. Some specific limits for the material properties are also given in Table 4.5. In addition to the listed required tests, there are three additional properties of repair materials that can be tested. But no specific limits are given for the three optional properties in ASTM C928: • • •
Flexural strength: ASTM C 78. Time of setting: ASTM C 403. Resistance of concrete to rapid freezing and thawing: ASTM C 666.
One can see that the above required tests are not very difficult to perform.
164
Conventional repair and strengthening
Table 4.5 Performance requirements of repair materials in ASTM C928 Materials’ performances
R1
R2
R3
Compressive strength 3 h: 500 (psi) 3 h: 1000 (psi) 3 h: 3000 (psi) 1 d: 2000 (psi) 1 d: 3000 (psi) 1 d: 5000 (psi) 7 d: 4000 (psi) 7 d: 4000 (psi) 7 d: 5000 (psi) 28 d: ≥7 d 28 d: ≥7 d 28 d: ≥7 d Slant shear bond 1 d: 1000 (psi) 7 d: 1500 (psi) strength Length change From 3 h to 28 d in water: ≤+0.15% From 3 h to 28 d in air: ≤−0.15% Scaling resistance of Max visual rating: 2.5/Max scaled material: 5 kg/m2 25 cycles Consistency of Concrete slump: Concrete slump: 3 (in). concrete or mortar 3 (in). Mortar flow: Mortar flow: 100% after 15 min after 100% after 15 min 5 min after addition of mixing liquid
Therefore, manufacturers of repair materials are currently using ASTM C928 as acceptance criteria of their products. If all the requirements of ASTM C928 are met, the repair material should meet the desired service life. However, premature damage of many repaired concrete structures has frequently been reported. This simply means that either the testing standard was not carried out accurately, or the required tests in ASTM C928 are not sufficient to characterize the durability properties of the repair materials. In the first case, better quality control of the testing process should be implemented, and in the second case, more advanced or more reliable testing methods should be developed in future to meet the need. There are several different bond test methods, in addition to ASTM C882, that may be used for various repair materials other than cementitious repair materials (ICRI 1997). These bond tests include ACI 503 R (Use of Epoxy Compounds with Concrete), the Michigan DOT Shear Bond Test, ASTM C 1042 (Bond Strength of Latex Systems Used with Concrete), and AASHTO T 237 (Testing Epoxy Resin Adhesive). Xi and Li (2004) made an extensive review of testing methods for concrete repair materials. British and European Standards for the protection and repair of reinforced concrete structures are under development (http:// www.concreterepair.org.uk/cra/StdsBSRPC.pdf). A comprehensive package of standards will be developed, including test methods for specific properties, specifications for important repair materials, coatings, mortars, bonding agents, and injection materials. The package of standards is drafted by Technical Committee 104 of the European Standards body, CEN. TC 104 is the European Standards committee responsible for specifications of concrete. Sub-committee 8 of TC 104 has drafted a series of standards for repair and protection of concrete structures, “The New Approach to Concrete
Conventional repair and strengthening
165
Table 4.6 European Standard EN1504: the new approach to concrete protection and repair EN 1504 Part 1 EN 1504 Part 2–7
EN 1504 Part 8 EN 1504 Part 9 EN 1504 Part 10
Definitions Products 2 Surface protection for concrete 3 Repair mortars, structural and non structural 4 Structural bonding materials 5 Concrete injection materials 6 Anchoring of reinforcing steel bar 7 Reinforcement corrosion protection Quality control and evaluation of conformity General principles for the use of products and systems Site application of products and systems and quality control of the works
Protection and Repair”. Table 4.6 lists all the parts of this standard, EN1504. 4.2.3 Laboratory testing methods for bond strength Of the five material properties required to be tested by ASTM C928 for cementitious repair materials, the bond strength is the one specifically important to repair materials, and it is also the most difficult one to measure accurately, thus we will discuss in detail the test methods for bond strength. There are many testing methods available for testing the bond strength of cementitious materials (Xi and Li 2004). The testing methods can be divided into five groups according to the type of stress carried by the interface (Halicka and Krol 1999). As shown in Figure 4.3, these five groups are tension, shear, shear and compression, torsion, and others (Saccani and Magnaghi 1999; Xiong et al. 2002; Soares and Tang 1998; Austin et al. 1999; Wall and Shrive 1998; Knab and Spring 1989; Yang and Zhu et al. 2000; Yang et al. 2000). In addition to the five groups, the ultrasonic test method (Liu et al. 1998) is also used to evaluate bond strength. Each testing method is influenced by different combinations of factors and on its own cannot give a comprehensive description of the required material properties. On the other hand, it is also important to know how to apply and interpret the results obtained from the bond tests to practical situations, where the interface may be in a different stress state from the tested stress state. Among the available test methods, shear, tension, and shear and compression are the frequently used methods for testing the performance of an interface between a repair material and a substrate. The other two types of testing methods are occasionally used in research laboratories or in the practice. The following is a brief discussion of the advantages and disadvantages of the three types of frequently used testing methods.
166
Conventional repair and strengthening
Figure 4.3 Bond strength test methods.
(A) SHEAR TESTS
Of all the available testing methods for the bond strength, the shear test is the most frequently used method for testing the performance of an interface between a repair material and a substrate under different stress states. If the bond surface is straight and smooth, such as a surface from a saw cut, the shear failure line can pass along the bond interface and cause a pure shear failure. If the bond surface is not straight and smooth, such as the bottom interface between a patch and substrate, there will be a strong mechanical interlock due to the surface texture, resulting in an incorrect bonding
Conventional repair and strengthening
167
capacity for the repair material. The insensitivity of the shear test to surface roughness can be overcome by testing in tension and torsion, to generate a combination of tensile and shear stresses. (B) COMPRESSION AND SHEAR TESTS
This test method was adopted by ASTM C882 (called the slant-shear test), in which the bond interface is under a combined stress state of compression and shear. The compressive slant-shear loading configuration is shown in Figure 4.4. This test is used by repair material manufacturers to evaluate the bond strength of repair materials. This test, however, has some shortcomings: • • •
Failure is crucially dependent on the angle of the plane. It is insensitive to surface roughness and condition, only producing bond failures with smooth surfaces (Robins and Austin 1995). The test is sensitive to the difference in elastic moduli of the repair and substrate materials.
Considering these shortcomings, some researchers have proved that the effect of surface roughness depends upon the inclination of the bond plane in the compression shear test, and there is a critical angle associated with a particular surface roughness (Austin et al. 1999). A critical bond angle can be defined as the inclination at which the load required to produce a bond failure is a minimum. Fortunately, the critical bond angle corresponding to smooth surface is about 30°, which exactly meets the specification of ASTM C882. In real repair projects, the concrete to be repaired is often sawed by a diamond saw and the interface of new and old concrete is quite smooth.
Figure 4.4 Compressive-shear loading configuration in ASTM C882.
168
Conventional repair and strengthening
Therefore, ASTM C882 is suitable for determining the bond strength. Moreover, the testing method is comparatively simple and easy to perform in practice. (C) TENSION TESTS
Tension tests are more difficult to perform than the compressive shear test and the shear test. Although the pull-off test (see Figure 4.3, the first row) is a more direct simulation of an actual repair situation than the slant shear test, significant variation in pull-off measurements within a specimen was observed. The main cause of the variability was a delamination at the edges between the patch and substrate due to the shrinkage in the repair mortar (Marosszeky et al. 1991). Flexural tests can also be categorized as tension tests. There are several different flexural testing methods. The one shown in the last of the first row of Figure 4.3 is the three-point bend test. Flexural tests can be considered an indirect testing method for measuring the interfacial bond strength of repair materials with old concrete (Yang et al. 2000), in which the bond property is evaluated by using a special beam made of new and old concretes. In the flexural test, the size effect on flexural bond strength is important and should be considered. Some researchers prefer to use flexural tests instead of the direct tensile test to evaluate the bond capacity, mainly because the direct tensile test is very difficult to perform. As illustrated in Figure 4.5, a fourpoint bending test was carried out on a beam with a prefabricated crack (a notch) at the bottom of the beam. The applied load and the crack mouth opening displacement (CMOD) were recorded, and then by analysing the load–CMOD curve, one can obtain the tension softening diagram (see Figure 4.5), which is the relationship between crack opening w and the stress σ in the interface of the new-old concrete (Kunieda et al. 2000). The σ–w relationship can be considered a measure of the bond strength, which is not a constant. In this sense, the bond strength measured by using other testing methods is actually an averaged value for the interface bond strength. It should be realized that the flexural testing methods are not standard test
Figure 4.5 Determination of tension softening diagram from the bending test.
Conventional repair and strengthening
169
methods for evaluating bond strength but sometimes used in research laboratories. 4.2.4 Field testing methods for bond strength Bond strength of repair materials has been measured both in the laboratory and in the field. Rizzo and Sobelman (1989) evaluated the selection criteria for repair materials. A number of adhesion/bond test methods were discussed, including direct tension, pull-off, direct shear, flexure, and slant shear. A compilation of studies of each method was summarized by Knab (1988). Several field test methods have been proposed to evaluate bond properties and the performance of repair materials in general. Tensile bond tests are gaining in popularity for field testing because of their relative simplicity and the ability to meet the requirements imposed on in-situ bond strength test. Tensile test methods can be divided into indirect and direct techniques. The pull-off test method is one of the tensile test methods. Unlike the other bond test methods that are used for laboratory testing, the pull-off test can be used in the field to evaluate the bond strength between repair material and parent concrete in a structure (Figure 4.6). The first modern development of the pull-off concept for strength testing of in-situ concrete was undertaken independently in the United Kingdom at Queens University, Belfast (Long and Murray 1984), and in Austria, where it was called tear-off test (Stehno and Mall 1977). This led to “Limpet” test equipment being
Figure 4.6 Pull-out tester made by Proceq. The dolly provides the connection between the loading device and the concrete surface.
170
Conventional repair and strengthening
commercially available in the United Kingdom. Further test equipment has since been developed in several countries, leading to a wide range of test configurations and procedures now being available (McLeish 1993). A number of different pull-off tests have been reviewed by CIRIA (McLeish 1993), the majority involving cutting of the repair material interface before applying a tensile load. Mathey and Knab (1991) studied the bond strength of concrete overlays by using in-situ uniaxial tensile tests (pull-off tests with partial coring). Two types of equipment were used in the tests: a hydraulic, uniaxial tensile test apparatus, which was a modification of the ACI 503R field-test apparatus, and a pneumatic apparatus developed at the National Institute of Standards and Technology (NIST). In the pull-off tests, cores were drilled through the overlay concrete to about 25 mm past the interface. A steel disk was then glued on the top surface of the core with a high-strength quick-setting epoxy. Then, a load can be applied to test for the bond strength. The important issue associated with pull-off tests is the depth of the core drilling into the existing concrete. It is suggested that the influence of the steel dolly and reaction tie on test results depend on the depth of coring into the substrate concrete. Ignoring the effect of drilling depth may be one of the main causes of difficulties in reproducing and comparing test results. It is very important to have a standard for the core depth beyond the repair–substrate interface for the pull-off bond test. The depth of core drilling below the interface should be a minimum of 25 mm (l in.) or one-half of the core diameters, whichever is larger. Although there are different testing methods and equipment for carrying out the pull-off tests, the general procedures can be described as follows: 1 2
3 4 5
Marking and preparing the test area. Partial coring into the existing substrate perpendicular to the repair surface. In some cases, partial coring is done around the attached loading disc. Attaching the disc to the core, using an epoxy resin. Attaching a loading frame to the disc. A frame around the disc provides the reaction force to the load. Pulling the disc until the specimens fails.
At the interface between the old and new concrete, the stress distribution is not uniformly tensile, because of the complicated failure surface, which is not perfectly perpendicular to the loading direction. As a result, the stress state is more complex than uniaxial tension, and there are stress concentrations in the interface. Therefore, most of the pull-off test results do not provide an accurate value for the tensile bond strength; they instead provide relative comparisons among the test data obtained from different types of interfaces. In many practical applications, the pull-off tests are conducted to determine if the bond strength between repair and concrete substrate meets the specified failure criteria.
Conventional repair and strengthening
171
Bond failure is usually classified by failure patterns as well as specified by critical bond strength values. There are several bond failure patterns. One is called adhesive failure, which is the failure in the repair–substrate interface. Another failure pattern is called cohesive failure, which includes the fracture failure within the repair material or within the substrate concrete. The other failure pattern is called adhesive/cohesive failure. It is a mixture of the first two types of failure patterns, partial adhesive and partial cohesive. In general, it is not desirable to have adhesive failure. In another words, it is not desirable for the failure to occur at the repair–substrate interface. Cohesive failure and adhesive/cohesive failure are acceptable, providing that the bond stress is equal to or greater than the specified bond strength. If failure occurs at the steel disc–repair interface, then the pull-off result represents minimum bond strength, and the test should be repeated if the strength is not acceptable. 4.2.5 Repair materials From the material component point of view, the repair materials used for concrete structures can be divided into two types: cementitious repair materials and polymer modified repair materials. In addition, some other materials, such as coatings, are also used for repair work. There are several commercial companies supplying general concrete materials, such as W.R. Grace and Master Builder in the US who provide various types of products for concrete repair work. There are also some companies specializing in materials for fast concrete repair, such as Conproco (http://conproco.com/ applicationCharts.htm) and Sika (www.sikausa.com). The websites and brochures developed by the material suppliers are good references for understanding basic properties of the repair materials. 4.2.5.1 Cementitious repair materials There are several types of cementitious materials that are often used for repair concrete structures, including Portland cement-based and gypsumbased concrete, magnesium phosphate concrete, and high alumina concretes. (A) PORTLAND CEMENT-BASED CONCRETE
Regular Portland cement concrete (PCC) is the most commonly used repair materials for spall repair. Usually, PCC used for patching should meet the following requirements: • • •
PCC (ASTM C 150), Type I and II, 362 kg/m3 (or higher); Maximum w/c ratio: 0.42; Total air content: 6 percent to 8 percent;
172 •
Conventional repair and strengthening Minimum compressive strength: 3500 psi (24 MPa) at 7 days.
Several different mix designs can be used for repair concrete. Perkins (1986) has suggested mix proportions of concrete and mortar for surface patching repair as follows: Concrete: 1 part ordinary or rapid hardening Portland cement; 2.5 parts concreting sand; 2.5 parts 10 mm coarse aggregate; The amount of mixing water should be such as to give a slump of 50 mm ± 25 mm which should be adequate for the compaction of small areas of concrete. Mortar: 1 part ordinary or rapid hardening Portland cement; 3 parts coarse concreting sand. To prevent excessive shrinkage, the total water content should be kept to a minimum. If an increased slump is desired to improve workability, a highrange water reducer should be added. High-range water reducers should not be used without first testing trial batches. If the repaired structure must be opened to use relatively quickly, rapid-setting or high early-strength materials, such as set-accelerator, may be used. Rapid set hydraulic cement/ mortar should meet the following requirement: •
• •
•
• •
Minimum of 30 minutes time to initial set as tested by ASTM C403, Standard Test Method for Time of Setting of Concrete Mixtures by Penetration Resistance. The mixture should have sufficient workability to allow placement and consolidation before initial set. Field-cured cylinders should have a minimum compressive strength of 1,200 psi (8.25 MPa) at 2 hours. Expansion/shrinkage should be tested in accordance with ASTM C157, Standard Test Method for Length Change of Hardened Hydraulic Cement Mortar and Concrete. The acceptance values for shrinkage are listed in Table 4.1, and the thermal expansion of concrete is discussed in section 4.2.1.1. Freeze–thaw durability should be tested in accordance with ASTM C666, Standard Test Method for Resistance of Concrete to Rapid Freezing and Thawing. The specimens should maintain at least 80 percent of their relative dynamic modulus of elasticity after 300 cycles of testing. The material shall not contain any other form of chloride. The product shall be used before its recommended shelf life expires.
Most structure owners want to reduce the period of repair. Therefore repair materials that can gain early strength have become very popular for many applications. Superplasticizer may be needed as an additive when the time
Conventional repair and strengthening
173
available for the repair is limited. For surfaces exposed to freeze–thaw/salt cycles, an appropriate air void content in the repair mortar or concrete must be achieved. It was shown by many researchers that very early strength (VES) concretes can be achieved by using conventional materials, normal mix designs, regular placing and curing practices. Two kinds of VES concrete (A and B) were developed by Virginia DOT. For VES A concrete, a minimum compressive strength of 2000 psi is reached 6 hours after water is added to the concrete mixture using Portland cement with a maximum water–cement ratio of 0.40. For VES B concrete, a minimum compressive strength of 2500 psi is reached 4 hours after water is added to the concrete mixture by using Pyrament PBC-XT cement, with a maximum water–cement ratio of 0.29. Both types of concrete can achieve a minimum durability factor of 80 percent at 300 cycles of freezing and thawing. Depending on the thickness of repair patches, mortar or concrete can be used for various repair work. The maximum aggregate size should not exceed 1/3 of the required thickness. For heavy traffic areas, a cement mortar or concrete with good bond should be used with a minimum thickness of 35 mm. For normal traffic area, this requirement reduces to a minimum 20 mm in thickness. Concrete mortar and concrete are applied by casting, trowelling and spraying. Casting is used for large repair jobs with horizontal and vertical surfaces. A possible problem with casting application is the adhesion to the old concrete. Spraying application of concrete is similar to shotcrete and will be discussed later. (B) GYPSUM-BASED CONCRETE
Gypsum-based (calcium sulfate) patching materials gain strength rapidly and can be used in situations where the ambient temperature is near freezing point (FHWA 1999). However, gypsum concrete does not appear to perform well when exposed to moisture or moist weather. In addition, the presence of free sulfates in the typical gypsum mixture may promote steel corrosion in reinforced concrete structures. (C) MAGNESIUM PHOSPHATE CONCRETE
Magnesium phosphate concrete sets very quickly, and therefore has high early-strength. It can be used to make impermeable patches that bond to clean and dry surfaces. Many research studies have been carried out on phosphate cement-based materials for rapid repair of concrete. Phosphate cement-based material usually is prepared by mixing the MgO (M) and NH4H2PO4 (P) powder with borax (B). Some researchers (Yang and Zhu et al. 2000; Abdelrazig et al. 1988; Seehra et al. 1993) studied the chemical compositions and mechanical properties of the materials, and their study
174
Conventional repair and strengthening
shows that MPB has good bond and compatibility with old concrete as well as low shrinkage. Yang (2002) studied the de-icer scaling resistance of MPB and found that MPB mortars have quite high de-icer frost resistance. However, these materials are extremely sensitive to water on the surface, and even a very small amount of extra water in the mix severely decreases its strength. They also cannot be used with limestone aggregates. MPB is a neutral or weak acidic material that may not protect the steel reinforcement from corrosion. In addition, this type of material contains a high level of Na+ ions, which may cause alkali–silica reaction in the concrete. (D) HIGH ALUMINA CONCRETE
Calcium aluminate concretes gain strength fast, bond well and shrink very little during curing (FHWA 1999). They are often used as refractory concrete. However, they may lose strength over time because of the chemical conversions that take place. 4.2.5.2 Polymer modified repair materials Polymer modified repair materials are a combination of polymer resin, aggregate, and a set initiator. Polymer modified repair materials are very useful not only for repairing concrete buildings but also for overlay applications on reinforcement concrete slabs and bridge decks mainly because they set very fast and have low shrinkage. Fast setting allows the repaired section of the structure to be used earlier, and low shrinkage ensures the durability of the repaired section. However, some of the polymer modified repair materials are sensitive to moisture. Magnat and Limbachiya (1997) studied three typical repair materials: high strength non-shrinkable concrete, a mineral-based cementitious material with no additives or coarse aggregate, and a cementitious mortar containing styrene acrylic polymer with fiber additives. The research results show that the shrinkage of polymer concrete was very sensitive to the relative humidity of the environment as compared to the other two concretes. The total shrinkage of the polymer mortar is the highest among the three repair materials despite the presence of some fiber additives. Furthermore, the compressive creep strain of polymer concrete is also the greatest as compared to the other materials. Moreover, there is significant discrepancy between the coefficient of the thermal expansion and the elasticity moduli of some polymer modified concretes and conventional Portland cement concrete. The difference in the coefficient of thermal expansion will generate a significant interface shear stress between polymer modified concrete and existing (regular) concrete upon temperature variations. The difference between the elasticity moduli will also generate interface shear stress between the new and old concrete. There are several types of polymer modified concretes commonly used
Conventional repair and strengthening
175
for repair applications, including epoxy, methyl methacrylate, and polyurethane modified concretes. (A) EPOXY CONCRETE
Epoxy usually has short curing time, high early strength, and high resistance to chemical attacks. Therefore, epoxy modified concretes or epoxy concretes are often used for repair work when a short curing time, a high early strength and a high resistance to chemical attack are required. Epoxy concretes are especially suitable for repair of surface damage and edge damage at joints. But epoxy resin systems have some disadvantages. As discussed earlier, they have larger coefficients of thermal expansion than that of concrete. Besides, they are relatively expensive. Resin systems usually have mechanical properties that differ from those of the concrete substrate, which result in the build-up of stresses at the interfaces as the resin cures. So a clean and sound substrate surface is very important for a successful surface repair (Saccani and Magnaghi 1999). Epoxy resins usually consist of two components, one is the active ingredient and the other the catalyst or hardener. It is important that the component materials of resin systems be mixed in the proportion suggested by the manufacturer. Epoxy resin repair materials should conform to ASTM C881, Standard Specification for Epoxy-resin-based Bonding Systems for Concrete. The commonly used mix design for epoxy mortar/concrete is four to seven parts dry silaceous aggregate to one part resin by weight. The maximum aggregate size should be less than 0.50 in. Aggregate passing the #100 sieve is generally excluded. The aggregate should preferably be under oven-dried condition. If not, its moisture content should not be greater than 0.5 percent. More information on the use of epoxy mortar/concrete can be obtained from ACI 546.1R, Guide for Repair of Concrete Bridge Superstructures, and ACI 503.4, Standard Specification for Repairing Concrete with Epoxy Mortars. Epoxy concretes are impermeable and excellent adhesives. They have a wide range of setting times, application temperature, strength, and bonding conditions. Epoxy gives off a large amount of heat during its hardening process, and thus, for deep epoxy repair work, the repair material should be placed in several lifts to control heat development in the concrete structure. (B) METHYL METHACRYLATE CONCRETE
Methyl methacrylate concrete has high compressive strengths and good adhesion (FHWA 1999). The binder consists of a high-molecular-weight methyl methacrylate (HMMA) monomer that is polymerized in place. Initiators and promoters are added to start polymerization and to accelerate polymerization, respectively. HMMA is preferred over other methyl meth-
176
Conventional repair and strengthening
acrylate monomers as it is less flammable. Other advantages of HMMA are low viscosity, high bond strength to concrete, and relatively low cost. The quantity of HMMA monomer to be mixed should be divided into two equal portions. The entire required quantity of initiator should be thoroughly mixed with half the monomer, and the entire required quantity of promoter mixed with the other half of the monomer. Then the two portions of monomer should be blended. The monomer should be mixed in clean, electrically grounded containers. Mixing should be conducted in a wellventilated area out of direct sunlight. A class B fire extinguisher should be available during mixing and placing. The aggregate should conform to ASTM C33. The maximum aggregate size should be less than 0.50 in. The aggregate should preferably be oven dry, but in no case should the moisture content exceed 0.50 percent. More information on the use of HMMA mortar/concrete for repair of bridge decks can be obtained from ACI 546.1R, Guide for Repair of Concrete Bridge Superstructures, Chapter 6(19). In addition, many methyl methacrylates are volatile and may pose a health hazard from prolonged exposure to the fumes. (C) POLYURETHANE CONCRETE
Polyurethane concretes generally consist of a two-part polyurethane resin mixed with aggregate. Polyurethanes generally set very quickly (FHWA 1999). Some manufacturers claim their materials are moisture-tolerant; that is they can be placed on a wet surface with no adverse effects. (D) LATEX MODIFIED CEMENT SYSTEMS
Latex is a dispersion of very small particles of an organic polymer in water. Latex modified cement systems should be proportioned in accordance with the specified mix design. The amount of water used in the mixture should be kept as low as possible to reduce shrinkage. Some research studies showed that the total water/cement ratio may fall within the range of 0.25–0.35 by weight (Meinheit and Monon 1984). To prevent latex losing its stability, airentraining agent should be avoided in the mixture to prevent the formation of small air bubbles in the mix. Latex modified concrete is often used for repairing bridge decks and parking garage deck overlays. The main drawback of latex modified cement system is that it is sensitive to high temperature (i.e. 29.4 °C or above) due to the film-forming characteristics of the latex. Besides, latex modified cement system is not suitable for underwater applications since it needs to dry in the air. The hardening mechanisms of the latex modified cement system are similar to regular concrete. During the curing process, the Portland cement hydrates as usual. In addition, a network of polymer film is formed within the mixture of aggregate and hydrates. This system provides high early strength, better flexural strength, better bond, better freeze–thaw durability,
Conventional repair and strengthening
177
lower permeability and better durability than those of regular concrete. For instance, the bond strength of latex modified cement system may exceed that of the unmodified cement system by factors of 2–3. The latex itself tends to seal the micropores in the cement system, resulting in a major reduction of permeability of cement system. Also, due to the dispersing effect of components in the latex combined with water; latex modified cement system has better workability than normal cement system. Immediately prior to applying latex modified cement systems for surface repair, the substrate concrete must be pre-wetted to a saturated surface dry (SSD) condition in order to delay film formation of the latex. As one can see from the above descriptions, latex actually has several different functions when added in the concrete mixture: (1) increasing tensile and flexural strengths of the concrete; (2) as a water-reducing agent (plasticizer) for concretes with lower water/cement ratios, improving workability of fresh concrete and decreasing shrinkage of hardened concrete; (3) improving the bond between the repair concrete and the existing concrete; and (4) reducing the permeability of the concrete by increasing its resistance to aggressive chemicals. Therefore, latex modified cement systems are usually used for repair jobs which require special properties that regular concrete does not have: high adhesion, low shrinkage, high durability, and low permeability. There are many different latex products available for repair applications. They have different requirements for curing procedure. In general, moist curing is required for latex modified cement systems in surface repair for 24 hours to let the cement hydrate properly, then the moist curing is usually followed by 72 hour of dry curing. Because of its excellent adhesion characteristics, latex modified mortar has no minimum thickness requirement in general concrete repair, provided that the size of sand particles in the mortar is small enough.
4.3 Repair techniques The word “concrete” has become a symbol of strength and durability in most people’s minds, but in reality, concrete structures come under attack from both natural and manmade forces. The rate of degradation of concrete depends on a number of factors; only some of them are controllable. Fundamental understanding of these factors is essential to determine how to repair the distressed structure. Some controllable factors that affect the rate of degradation of concrete are quality of concrete placement, mix design of concrete, drainage condition of the structure, application of de-icing chemicals, etc. Non-controllable factors include the environmental conditions, such as moisture fluctuations, freeze–thaw cycles, and chemicals in surrounding water sources. Man-made forces affecting degradation of concrete generally fall into two categories: (1) design deficiencies or neglect; and (2) overuse and overload of the structures.
178
Conventional repair and strengthening
There are several types of concrete degradation that can be found in many concrete structures, which will be discussed in detail in later sections. One of the common degradations of concrete structures is the damage caused by corrosion of the embedded reinforcing steel. The corrosion of embedded steel reduces the cross-section area of the steel and thus reduces the load-carrying capacity of the structure. At the same time, corrosion results in various types of damage in the concrete surrounding the reinforcing steel. The damages include cracking, delamination, and spalling. Because reinforcing steel is virtually found in all reinforced concrete structures, corrosion of reinforcing steel is a very critical and widespread problem. The corrosion of steel and the deterioration of concrete are two individual durability problems for reinforced concrete structures and they tend to occur together. The accumulation of rust formed in the corrosion process of a reinforcing bar results in cracks in the surrounding concrete cover. The first crack may be very small but it invites easier intrusion of moisture and aggressive chemicals such as chloride ions from the de-icing salts, in turn, the combination of the moisture and the chemicals accelerates the corrosion process. Eventually and inevitably, the outward symptoms of scaling, cracking and spalling gradually begin to appear in concrete structures. Figure 4.7 shows a bridge pier experiencing severe corrosion damage.
Figure 4.7 Example of steel corrosion.
Conventional repair and strengthening
179
The most important task in planning a successful repair of concrete structures is to find the cause (damage mechanism) of the deterioration of the structure and apply effective methods to stop the damage mechanism. This, apparently, is a very difficult task even for experienced professionals. In the following sections of this chapter, we will first introduce general methods that have been used to repair concrete structures such as how to repair concrete surface damage and how to repair concrete cracking, without going into details on the cause of the damage in the concrete. Then, we will further discuss in greater depth several commonly seen deteriorations in concrete and the specific repair methods for the damage, including fire damage, corrosion damage and the damage related to alkali–silica reaction in concrete. Finally, we will discuss some specific repair methods for special structures such as bridges and pavements, which are different from buildings. Concrete repair provides extended service life to the structures by integrating new materials with existing materials to form a composite structure that can withstand environmental and mechanical loadings. There are many advanced solutions, far beyond the simple concrete patch, that can be utilized to implement an effective concrete repair. Basic understanding of concrete repair options, such as surface repair, protection, stabilization, strengthening and waterproofing, will successfully provide extended service life of the structure (Emmons 2006). It is important to mention that, in addition to the commonly used repair methods, some of the repair methods described here may not be commonly used methods and they are still in the experimental or developmental stage. For these repair methods, the readers should obtain more detailed information or consult experienced professionals before actual implementation. 4.3.1 General repair strategies In repairing deteriorated concrete structures, the primary object is to restore the concrete structural member or the whole structure to its original shape and condition, to ensure structural integrity, durability, and composite behavior and, sometimes, to match the existing concrete in color and appearance. Repair is not only to recover the original condition of an infrastructure, but also to extend its service life to satisfy the design requirement. A successful repair is not only involved in the portion of the structure to be repaired, but also the reinforced members connected to the repaired portion. Basically, the repair procedure for reinforced concrete structures consists of some or all of the following four steps: 1
2
Removal of damaged concrete to expose all damaged portions of the structure. The damaged concrete portion may contain cracks, delamination, and etc. If there is reinforcing bars in the damaged portion of the structure, the surface of the steel should be cleaned to remove any rusting layers, and
180
3
4
Conventional repair and strengthening additional steel should be incorporated if necessary. A protective coating may also be applied to the existing and new steel surfaces. Application of repair mortar or concrete to replace removed concrete, which serves two purposes: (a) the repair material prevents the ingress of corrosive agents (physical protection); and (b) to re-passivate the steel (chemical protection). Primer may need to be applied first to improve bond between existing concrete and repair material. To enhance protection for steel, external membranes may be applied over the repaired section, or the whole concrete surface.
Like the selection of a proper repair material, the selection of a repair method is often a difficult process which involves consideration of a large number of factors, some of which are technical, some economic, and others may be practical. A successful and durable concrete repair can result only from correct choices for repair method and for repair materials. In choosing an appropriate repair material for concrete structures, there are a number of considerations such as those discussed in earlier sections: strength, durability, compatibility, ease and safety of application, and last but not the least, cost. Cost is a very important issue. In a limiting case, the cost of repair should not be more than the cost of replacement. But in some cases, even if the cost of repair is more than the cost of new construction, people preferred to choose repair work because of other considerations such as environmental preservation and historic restoration for some landmark buildings. Repair work not requiring strengthening of the structure may be classified in different ways: either by geometric feature of the area to be repaired such as vertical and horizontal surfaces and cracks or by the failure mechanisms of the structure or structural member such as fire damage and corrosion damage. For the first type of classification: the repair methods for concrete surfaces and concrete cracks will be described. For the second type of classification, the repair methods for fire damage, corrosion damage, and ASR damage will be discussed. The following sections will be presented in the remaining part of this chapter: 1 2 3 4 5
Repair of the concrete surface Repair of concrete cracks Repair of fire damaged concrete Repair of corrosion damage Repair of ASR-induced damage.
4.3.2 Preparation methods Depending on the deterioration mechanisms and damage conditions of concrete surfaces, different methods may be available for repair. For shallow and minor damage in the concrete surface, surface coating, surface
Conventional repair and strengthening
181
impregnation and surface sealers may be applicable. For deep and serious damage, the damaged and deteriorated concrete should be removed and replaced by new repair material. For all cases, repair work begins with a proper surface preparation of the damaged or deteriorated concrete surface to ensure good bonding development between the new material and the old material. Many surface cleaning and preparation procedures are available and the choice will depend on the condition of the substrate and the degree and nature of contamination. The ability of repair materials to wet out the concrete surface varies, and consequently, adhesion to concrete relies greatly on the mechanical bond. 4.3.2.1 Preparation of the surface In all cases where concrete surfaces are repaired, the condition of the existing concrete in the exposed damaged area is of primary importance for the durability of the repair. The repair work can seriously be compromised if the adhesion between the new repair material and the existing sound concrete surface is not strong. Surface preparation includes all the steps taken after removal of large amounts of deteriorated concrete. It also includes the steps taken to prepare surfaces on which the new repair material is to be placed, even if no concrete is removed. In order to fully understand the extent of damage to the structure, all damaged concrete should be removed. Concrete removal techniques should be selected on the basis of safety, economy and their effect on the remaining sound concrete. They also depend on the situation, especially on the extent and thickness of the layer which has to be removed, as well as on the type, location and position of the damage in the structure (FIP 1991). There are many industry guides and standards related to surface preparation for repair. The guides and standards that should be referenced, include but not limited to the following: ASTM D 4258 ASTM D 4259 ASTM D 4260 ASTM D 4261
Surface Cleaning Concrete for Coating Abrading Concrete Acid Etching Concrete Surface Cleaning Concrete Unit Masonry for Coating ASTM D 5295 Guide for Preparation of Concrete Surfaces for Adhered (Bonded) Membrane Waterproofing System ACI 546R-04 Concrete Repair Guide ACI 224.1R-07 Causes, Evaluation, and Repair of Cracks in Concrete Structures NACE RP0591-91 Coatings for Concrete Surfaces in Non-Immersion and Atmospheric Service NACE TCR6G166 Surface Preparation of Concrete Coatings
182
Conventional repair and strengthening
For different types of damage, there are different technical guidelines for surface preparation. For example, ICRI guideline 03730 (ICRI 2002) covers surface preparation for repair work of corrosion damage. Surface preparation should introduce as little damage as possible to the concrete remaining in place. On the other hand, it is difficult to determine whether all the damaged material has been removed, because the zones of damaged or deteriorated concrete are usually difficult to define. Surface imperfections should be well treated during the surface preparation process. For instance, sharp edges should be ground smooth and blowholes should be filled with a suitable filler. Proper surface preparation provides a dry, even and level surface free of dirt, dust, oil and grease. Removal of surface contaminants allows primers and repair materials to have direct contact with the substrate, increasing the surface area of high quality existing concrete, increasing roughness of the surface, and providing increased anchorage of the applied new repair material. Table 4.7 lists some of commonly used surface preparation methods. In order to choose the best concrete removal method or combination of methods, the following safety, environmental and job-related information should be considered in addition to the industry guides and standards: 1
2
The cleaning and surface preparation instructions supplied by the material manufactures must be carefully studied. Some specific surface preparation methods are required for special repair materials. The location of repair work, e.g. inside a building or outside. The
Table 4.7 Standards of surface preparation Surface conditions
Methods
Dry and free from grease, free from dust, flakes, salts and laitance, with a sharp even finish
Application of emulsified 2.5–3.0 N mm2 degreaser and washing clean with water, grit blasting, highpressure water jetting and acid etching Application of emulsified 1.5–2.0 N mm2 degreaser and washing clean with water scraping off any loose matter, power wire brushing and vacuum cleaning to remove dust Application of emulsified 0.5–1.0 N mm2 degreaser, washing clean with water, scraping and hand wire brushing to remove loose matter and brushing clean
Dry and free from grease, free from dust, flakes, salts and superficial laitance
Dry and free from grease, free from dust flakes and salts
Expected tensile adhesion
Conventional repair and strengthening
3
4
183
restrictions with respect to noise, dust, vibration, exhaust fumes and disposal of wastewater should be considered. This is important especially when the repair work is inside a building. The additional load due to construction equipment and construction materials. This is important when a large amount of repair materials and equipment must be piled up temporarily during the repair process. It is crucial to make sure there is no overload applied to the structure. The thickness of the damaged concrete to be removed. This is important when choosing the concrete removal equipment.
(A) MECHANICAL METHODS
A wide variety of mechanical equipment is available for the surface preparation of concrete, such as pneumatic drill, electrical saw, electrical hammer, compressed-air hammer, mill machine, sand blasting, and so on. In general, these devices are of two types: rotary and impact. Rotary equipment includes rotary discs and grinders which are usually used on the substrate containing concrete of relatively low compressive strength that does not have a steel trowelled finish. They are not effective on hard, dense concrete. Abrasive blasting is a fast and effective method of surface preparation using rotary equipment. A widely used abrasive tool is copper slag grit. The grit is propelled against the concrete surface using large volumes of compressed air. But the aggressive abrasion may cause numerous blowholes which have to be filled before repair work. Impact tools such as electric hammers, scabblers, and needle guns will effectively remove several millimeters of concrete. Scabblers use compressed air to hammer piston-mounted bits into the concrete surface. They tend to roughen the concrete surface more than either abrasive blasting or shotblasting. Scabbling operations are dusty, noisy, and produce some vibration, so vacuuming or spray-pressured water is always required to improve the working condition. Impact tools pulverize the concrete and can cause fracture of the concrete below the surface. Consequently, it may be necessary to use another means (such as water jetting or wet sandblasting) for the final cleaning of the surface. Heavy hand labor is needed when using mechanical methods, such as hammer and hand tools to prepare the surface for repair. If a concrete layer with a small thickness is to be removed, chipping-off is recommended. A milling device is helpful in removal of concrete to a flat surface. Several milling passes may be required for a thick layer of concrete. Sand-blasting is suitable for roughening of concrete surface, but it is not suitable for removal of a thick concrete layer. For cleaning concrete surfaces, high pressure water jetting are often used, but in case of heavy pollution or ingrained dirt, it may be necessary to use wet sand blasting and manual scrubbing with detergents to clean the concrete surface. A power wire bush is quite often used to clean the concrete
184
Conventional repair and strengthening
surface. Concrete treated by this technique exhibits a regular and relatively blowhole-free surface. A hydraulic water jet with 10–40 MPa pressure is able to remove loose concrete particles. A great advantage of this technique is that there is no dust produced. A hydraulic water jet with a higher pressure up to 40–120 MPa is an efficient way to remove areas of soft concrete surface. With a higher pressure to 140–240 MPa, the hydraulic water jet may be used to cut concrete. Whenever deteriorated concrete is removed using impact tools, the surface of the remaining concrete may be damaged. If this damaged layer is not removed, the repair material may debond from the substrate. Therefore, the remaining concrete should be further prepared using wet sandblasting or high pressure water jetting to remove this damaged surface material. Usually the removal of limited areas of concrete to permit a repair requires the sawcutting of the perimeter of the areas to minimize further damage to the area to be repaired. (B) CHEMICAL METHODS
Many deterioration processes result from penetration of the concrete by gases or salts in solution. In such cases, surface treatments are designed to reduce the passage of potentially deleterious substances and slow down the rates of deterioration. Concrete that is contaminated with oil, grease, or dirt requires chemical cleaning piror to the application of repair materials. Acid and alkalines are suitable for this type of work, for instance, acid etching is a particularly efficient method of surface preparation. Acid etching is especially suitable for areas demanding clean conditions since no dust is produced, but its most effective use is restrticted to floors due to practical difficulties with vertical or inclined surfaces (Allen et al. 1993). A 10 percent solution of muriatic (hydrochloric) acid is normally used for concrete surface cleaning in acid etching. The concrete should be pre-wet, and all oil, grease, paint, sealers, gum, tar and other foreign materials must be removed before etching. Chemical etching should be followed by vigorous scrubbing and thorough rinsing with water to remove all residues. If necessary, the surface should be water-jetted to ensure that all water-soluble salts have been removed. If thorough cleaning cannot be done after chemical etching, a neutralizing agent can be used. In this case, the neutralizing agent should be dispersed in water and then the solution can be used on the concrete surface. Diluted ammonia (4 percent solution) in water mixed at a rate of 0.9l concentrate to 19l water flooded over the surface of the concrete will be sufficient (Holl and O’Connor 1997). It is normally good practice to check the pH level of the prepared concrete surface. If the pH level is below 10, it is recommended to carefully wash the concrete surface with a weak solution of sodium hydroxide to raise the pH
Conventional repair and strengthening
185
level above 11. Conversely, if concrete containing reactive aggregates has been attacked by alkalis, the surface can be washed with water to lower the pH level below 12 (Newman 2001). It should be noted that muriatic acid, commonly used to etch concrete surface is relatively ineffective in removing grease or oil (CSA 1989). Besides, due to the potential risk of corrosion of the rebars in concrete, acids should not be used for reinforced and prestressed concrete (FIP 1991). (C) FLAME CLEANING
Flame cleaning is generally used to clean concrete surfaces that are to receive coatings or resinous overlays. This method is particularly useful for oilstained floors because it permits the application of coatings to the concrete immediately below. A special multi-flame oxy-acetylene blowpipe is passed over the concrete surface at uniform speed. The thickness of the concrete layer removed depends on the speed at which the blowpipe is moved and the properties of the concrete. The most suitable blowpipe speed lies between 0.02 m/s (0.066 ft/s) and 0.03 m/s (0.099 ft/s). Concrete and coating removal involves both the spalling and melting off of the surface. The laitance layer is usually removed to a depth of 1 or 2 mm (0.04″ or 0.08″) and in a few instances up to 4 mm (0.16″). The moisture content of the concrete has the greatest effect on the concrete removal – completely dry slabs do not produce much spalling, while slabs soaked in water prior to flame cleaning produce uniform concrete removal. European experience indicates that flame cleaning does not promote the migration of deep-seated oil to the surface, does not remove the alkalinity of the matrix – the surface gradually attains alkalinity similar to that of new concrete – and does not promote the development of any visible cracks in the surface. The method has proved useful for such applications as the recoating of concrete floors or the removal of defective elastomeric waterproofing membranes from parking decks (Mailvaganam et al. 1998). 4.3.2.2 Concrete surface requirement The desired condition of the concrete surface depends on the type of repair and the condition of the substrate. After the surface preparation, the strength of adhesion and moisture content of the remained concrete have to be evaluated to ensure a good surface repair quality. The strength evaluation can be done by pull-out test. In this method, a 50 mm diameter steel plate at least 10 mm thick is glued to the concrete surface after a corresponding circular-core drill cut is made in the concrete surface. After the glue hardens, the steel plate is pulled off the concrete surface and the tensile stress achieved in the pull-off test is referred to as the strength of adhesion. It is required that the measured strength of adhesion should be greater than 1.5 MPa. If the
186
Conventional repair and strengthening
adhesion strength is less than 1.5 MPa, the concrete is required to be removed to a further depth. The moisture content can be evaluated by mass measurement method by using a moisture meter, or using the method specified in ASTM D 4263, Indicating Moisture in Concrete by the Plastic Sheet Method. In this method, a concrete specimen is sealed by a plastic sheet about 24 in2 (15,500 mm2). If the concrete appears dark, damp, or wet, the presence of excessive moisture in the concrete is indicated. The concrete should be allowed more time to dry until the excessive moisture is removed (Holl and O’Connor 1997). When electrode-type moisture meters are utilized, care must be exercised, if readings are taken on surfaces recently wetted, since the concrete interior could be dry. It is wise to chip off some of the surface and repeat the test on the underlying surface. Normally, readings of less than 5 percent moisture content by mass are required before coating, and specific requirement by local building codes should be referred to regarding the residual moisture content before coating. The required moisture condition can be achieved by water spray or air dry out. The permissible moisture content depends on the selected repair method and the applied materials. A cement bond system requires wetting of the concrete surface, so that the existing concrete will not absorb water from a repair coating. However, an excessive water content may also be detrimental to adhesion. Therefore, the concrete surface should be moist but not saturated. For a plastic bond system, a dry surface is required, which means the moisture content is less than 6 percent by weight to a depth of approximately 20 mm. Good adhesion cannot be achieved if the water in old concrete prevents the penetration of liquid repair materials. Surface preparation also needs the removal of reaction products such as laitance that cover the surface by various methods described above. The removal of the surface contaminates allows primers, and the repair materials themselves, to have a strong bond to the substrate. For most inorganic materials, the concrete should be saturated and surface dry to prevent rapid loss of water from the repair material to the substrate so that shrinkage and cracking in the repair material are prevented. Maximum adhesion for organic (resin-based) materials is achieved by ensuring that the concrete surface is dry. For reinforced concrete, surface repairs may include proper preparation of the surface of reinforcing steel to ensure good bond development with the replacement concrete so that the desired behavior in the structure is obtained. The deteriorated concrete surrounding the reinforcing steel should be removed so that the original reinforcing steel is exposed and then cleaned. A pachometer may be needed to determine the depth of the steel in the concrete. A pachometer is an electromagnetic device used for determining the location and cover of reinforcement bars in concrete.
Conventional repair and strengthening
187
4.3.2.3 Bonding agents There are mainly two kinds of bond between the old concrete and new concrete: (1) physical bond through adhesion and cohesion; and (2) chemical bond through reactions among the materials in contact in the interface. In most cases, both types of bonding exist at the same time. Although adequate concrete bond strength can be achieved when patching, mortar or repair concrete is placed against a prepared surface that is water-saturated but surface dry, a bonding agent is recommended to use in repair work to improve the bond between the old and new concrete. It is essential to obtain the best possible bond at the interface with the help of bonding agents. Bonding agents may increase the cost of repair. The commonly used bonding agents for surface repair of concrete include: 1
2 3
4
Cement paste with relative low water–cement ratio. The cement paste can be brushed onto the surface to be repaired. Cement-based bonding agents usually require some initial protection (i.e. proper curing), because a rapid drying of the materials could lead to shrinkage cracking and delamination. The bonding agents should be applied soon after surface preparation is completed. Cement mortars with a cement and sand ratio of 1.0. Proper curing is also needed for this type of bonding agents. Polymer modified cements, such as latex modified mortars or pastes. In this type of bonding agent, the polymers are mixed with the cement paste or cement mortar via the mixing water. Additives, such as dispersion agents and emulsions, may be added to the mixture to improve the bond strength, workability and water retention capacity. Latex modified mortars have increased workability characteristics and a tendency to entrain air in the mix. Therefore, they should be mixed by a slow speed mixer or by hand to reduce the amount of air entrainment. With latex materials, it is advisable to wet the concrete surface beforehand, and allow the surface to become dry prior to installation. Resin systems, such as epoxy resin and hardener with or without fillers. Resin-based bonding agents do not generally need any protection during their curing period. In the case of epoxies, however, the concrete surface should be dry to allow the epoxy to penetrate into the surface pores. Damp concrete surfaces inhibit the proper adhesion of epoxy material and may resulting in debonding (Mailvaganam 1992). In many cases, filling materials are used with resins mainly because they are less expensive than epoxy resin, they can reduce shrinkage as the resin cures, and they can reduce temperature as the exothermic reaction takes place when the active ingredients of resins and the hardener are mixed together.
188
Conventional repair and strengthening
4.3.3 Repair of concrete surface 4.3.3.1 Shotcrete Repair of concrete surface normally uses shotcrete or gunite, and, sometimes, this technique is also referred to as sprayed concrete. Sprayed concrete is a mixture of cement, aggregate and water projected at high velocity from a nozzle by pneumatic pressure into a location in an existing structure or formwork where it is compacted by its own velocity to produce a dense homogeneous mass (Allen et al. 1993). The Concrete Society of the UK (1979, 1980 and 1981) has defined gunite as a sprayed concrete with a maximum aggregate size of less than 10 mm; whereas shotcrete is defined as a sprayed concrete where the maximum aggregate size is 10 mm or greater. This technique was first developed as long ago as the 1930s and is also suitable for concrete replacement and for strengthening structural elements. Within the category of shotcrete, a distinction can be made between a dry mix process and a wet mix process. In the dry mix process, the solid constituents, cement and aggregate, are mixed without the addition of water. This dry mixture is then delivered along the hose to the discharge nozzle where water is added in sufficient quantity to hydrate the mixture and to provide the right consistency so that the produced mixture can be projected at high velocity onto the structure to be repaired. The impact due to the high velocity compacts the mixture to form the required quality of the concrete. The dry mix process can produce sprayed concrete with very low water– cement ratios and almost no slump that enable it to be placed to a limited thickness on vertical and overhead surfaces. In the wet mix process, all the ingredients including the water, are batched and mixed and then pumped along flexible hoses into a discharge nozzle. At the nozzle, compressed air is introduced and projects the mixture into position at high velocity as shown in Figure 4.8. Nowadays, the wet mix process has become more and more popular because the working conditions are better than that for dry mix process and there is better quality control. When sprayed concrete for surface repair is used, in either a dry or wet mix process, pre-treatment of the old concrete surface is important. Sandblasting is efficient for surface treatment and the existing concrete surface should also be pre-moistened. If reinforcement steels are exposed after removing the deteriorated concrete, the rust on the steel must be removed and the steel must be cleaned down to the bare metal. Bonding agents, such as epoxy bonding agents, latex-modified cement slurries, or pure cement slurries can be used as a base coating to improve the bond between the sprayed concrete and the existing concrete substrate. But there are some arguments that there is no need for these bonding agents since they may be removed from the concrete surface during the initial pass of spraying concrete. Besides, during the initial application of spraying concrete, rebound is
Conventional repair and strengthening
189
Figure 4.8 Application of shotcrete for repairing a concrete arch.
high, resulting in a cement-rich layer left, which may fulfill the same function as the bonding agents (Crom 1986). One important note in the application of shotcrete is that the thickness of the concrete layer should be considered. If the thickness is large, multi-layer sprayed concrete is needed, which requires the preceding layer to achieve a sufficient hardness. If the thickness is more than 50 mm, minimum reinforcement is required. Evaporation protection is helpful in curing sprayed concrete after application to prevent a rapid drying out. The purpose of using sprayed concrete in surface repairs is to provide a relatively thin, high quality, protective layer on the area to be repaired. If prepared and applied properly, sprayed concrete can provide excellent bond, high strength, and a protective cover of low permeability to the existing steel reinforcement and concrete. Each process, both dry mix and wet mix, has its own advantages and disadvantages, and the choice of process depends on the cost and availability of equipment and crews, operational features and the particular circumstances of the application. The major advantages of both processes are that no formwork is needed for the placement of the concrete and that a variety of shapes of surface can be formed. The repair of concrete by shotcrete has very wide applications for thin walls, certain roofs, slabs, reinforced concrete tanks, tunnels, canals, coatings on brick, steel and masonry, encasement of steel for fireproof and repairs to
190
Conventional repair and strengthening
earthquake damages. The disadvantages for sprayed concrete include: the quality of sprayed concrete is largely dependent on the skill of the operator; the uniformity of the sprayed concrete is not good; the placing process of sprayed concrete is relatively dusty; it is generally more suitable for thinlayer repair, not good for thick section; and the control and inspection of the material and repair work by sprayed concrete are difficult. The dry mix process produces more dust and rebound than the wet mix process. Therefore, the wet mix process is favored for work in a confined environment such as tunnels and in environmentally sensitive areas. Since no water is added to the mixture until just before placing, the dry mix process is appropriate where sprayed concrete is to be used intermittently and spraying can be stopped and started at will. On the other hand, the wet mix process is more suitable for continuous spraying and it is often used for large-scale repair project, typically in new construction work. Most commonly used mixes for the dry mix process have an aggregate– cement ratio in the range of 3.5 : 1 to 4 : 1 by weight and the sprayed concrete in place shows good density, high strength (typically 40–50 N/mm2) and very good bond to a suitable substrate. In the wet mix process, cement contents are usually in the range of 350–450 Kg/m3 with a water–cement ratio down to 0.4. The mix, including cement contents, aggregate–cement ratio, maximum aggregate size and grading, has to be designed to be pumpable. Normally higher water–cement ratios are found in the wet mix process than those in dry mix process in order to achieve pumpability. Compared with the sprayed concrete made by the dry mix process, the in-place strength of the wet mix process sprayed concrete is lower, typically 20–40 N/mm2. In some cases, no-fines concretes needs to be used. No-fines concrete is a concrete that only contains normal Portland cement, water and coarse aggregate. The use of non-sand concrete has expanded into many areas such as: drainage pipes, wells, small retaining walls, pavements (for local roads and parking lots), and repair work. One of the disadvantages of such concrete is the difficulty in mixing and placing (pumping) due to the stiff consistency of the mixture caused by the low water–cement ratio required to prevent segregation. It was shown that non-sand concrete can be made pumpable by adding several admixtures (Amparano and Xi 1998). Pumping characteristics are based on the mixture’s ability to provide good slump and excellent cohesiveness. The purpose of the research was to obtain a pumpable nonsand concrete mixture, within the specified design parameters, which achieved two goals, that is, exhibits excellent pumping characteristics and yields equal or higher strength values than that of mixtures made without admixtures. Their results indicated very good performance of mixtures containing silica fume concurrently with the hydroxyehtyl cellulose admixture for non-sand concrete mixtures with 12.7 mm and 6.35 mm maximum size aggregates.
Conventional repair and strengthening
191
4.3.3.2 Replacement When the damaged concrete is in a large volume, especially when the thickness of the damaged concrete is large, the damaged concrete may be replaced by new cast of repair concrete. Strictly speaking, the shotcrete described in the previous section is also a means of replacing damaged concrete. It is a special repair method that can be used for large areas of damaged concrete with a small thickness, and it is basically used to repair surface damage on a concrete structure. When damaged old concrete is intended to be replaced, the compatibility between old and new concrete should be taken into account. The surface of the old concrete requires adequate preparation, careful cleaning and premoistening. Placing the new concrete should avoid the entrapment of air. Therefore, the formwork must be sufficiently rigid and tightly fitted to the existing concrete in a manner to minimize leakage of cement paste. The new concrete should be re-compacted to improve contact to the old concrete before initial setting. For large-volume replacement of old concrete, sometimes minimizing the temperature difference between old and new concrete may require special procedures (cooling of new concrete and/or heating of old concrete). In general, the replacement of concrete should have final properties, such as strength and modulus of elasticity, that match the existing concrete as closely as possible. Thus, the mix design of the repair concrete, such as type of cement, cement content and the water–cement ratio should be carefully chosen. 4.3.3.3 Grinding Surface grinding can be used to repair some bulges, offsets, and other irregularities that exceed the desired surface tolerances. This method is often used to repair concrete driveways, sidewalks, and roadways when a concrete slab is not levelled very well with the surrounding slabs. Excessive surface grinding, however, may result in weakening of the concrete surface and exposure of easily removed aggregate particles. If the concrete slab or the concrete member is to carry a heavy load, then the load-carrying capacity of the member should be examined with regard to the grinding process to make sure that the load-carrying capacity of the reduced cross-section is sufficient for the load. For these reasons, surface grinding should be performed subject to the following limitations: •
•
Grinding of surfaces subject to cavitation erosion (hydraulic surfaces subject to flow velocities exceeding 40 feet per second) should be limited in depth so that no aggregate particles more than 1/16 inch in crosssection are exposed at the finished surface. Grinding of surfaces exposed to public view should be limited in depth so that no aggregate particles more than ¼ inch in cross-section are exposed at the finished surface.
192 •
Conventional repair and strengthening In no event should surface grinding result in exposure of aggregate of more than one-half the diameter of the maximum size aggregate.
4.3.3.4 Surface patching Patching refers to the restoration of relatively small areas of damaged concrete to the profile of the surrounding concrete. This technique has been used for many years and is typically used to restore concrete sections damaged by delaminating and spalls. It is also an appropriate repair method when the concentration of aggressive chemicals is low to moderate and when the areas of visible deterioration are relatively few (Newman 2001). The disadvantage of surface patching is that the appearance of the repaired surface often becomes a problem if too many small patches are installed, as shown in Figure 4.9. Similar to other repair techniques, the surface preparation of the substrate is critical to a successful patching. The boundaries of the areas requiring patching should be clearly defined with a saw cut to a depth of about 6 mm and contaminated and unsound concrete within this area should then be removed. For easy and proper patching, the patch geometry should be simple, with minimal edge length and perimeter. Disc cutting at right angles to the surface (to a depth of 5–25 mm) is usually employed to achieve the required geometry. Feather edges should be avoided when cementitious
Figure 4.9 Many small patches applied on a bridge deck.
Conventional repair and strengthening
193
repair are used, but resin-based materials, such as epoxies which have considerably better bonding characteristics, can be feather-edged. The repair systems normally include bonding agents. The appropriate bonding agents include sand/cement, cement/latex slurries and epoxies. These bonding agents help in reducing loss of moisture from the repair and also improve the bond between the new and old concrete. Certainly, the use of bonding agents increases the cost of repair, with cement slurry being the cheapest and epoxy the most expensive. The bonding agent should be applied soon after surface preparation is completed. The repair system may also include reinforcement primer, polymer modified mortar, pore filler levelling mortar and protective coatings. When choosing repair materials, normally the mechanical and transport properties of the repair material should resemble as closely as possible those of the surrounding concrete. Portland cement mortars and grouts are the most frequently used materials for surface patching because of their good compatibility with the existing concrete. Latex modified mortar and epoxy mortar are often substituted for cementitious patching materials when a fast cure time, higher bond strengths, and feather edging are required. The addition of polymers to repair mortars improves their properties in different stages. For the mix design of the repair concrete, when polymers are used, the water dosage needed can be reduced, and thus also does the long-term shrinkage of the repair mortar. For the interface bond, the polymers help to improve the bond between the repair mortar and the substrate, and they can increase flexural and tensile strengths of the modified mortars and reduce permeability to the diffusion of moisture and carbon dioxide (Kay 1992). Rapid setting materials based on magnesium phosphate and high alumina cement are also used to patch damaged flat structures, especially when rapid temporary repairs are necessary. The current process to prepare repair materials is that all of the repair materials are provided in a packaged form as a system from one manufacturer. As far as mortars are concerned, preblended dry sand and cement may be supplied in one package, and the latex and the appropriate quantity of water in another. The ingredients are simply mixed together at the construction site to produce the repair mortar. There are several steps in the patching repair process: 1
Cut out the area to be repaired. The first step of a surface patching repair is to cut a groove around the anticipated edges of the repair. The edges of deteriorated areas should be cut perpendicular to the surface to a depth of at least 10 mm (Mailvaganam 1992). The shape of the repair should be delineated by straight lines and abrupt changes in width and depth should be avoided. The main surface-breaking work is usually undertaken by pneumatic or electric tools. Cutting by high pressure water jet is an alternative for surface breaking which causes less damage to the substrate and less noise. When heavy tools are used for surface preparation, the damage in the form of loosened aggregates should be
194
2
Conventional repair and strengthening removed with lighter tools. In the case of reinforced concrete with corroded reinforcement steel, concrete is usually cut out beyond the reinforcement steel to give a clear gap of approximately 20 mm below the steel bar, so that the backs of the reinforcement bars can be properly cleaned during the preparation of surface patching repair. These exposed reinforcement bars are then cleaned by grit blasting to satisfy the specified qualities of surface preparation. Sometimes, grit blasting followed by washing with fresh water and further blasting is needed in order to get a clean surface for patching. In severe cases it may be necessary to cut out the corroded reinforcement steel and install a new bar as a replacement. The new steel bar should be grit blasted to the same high standards as the existing reinforcement within the repair. Once the reinforcing steel has been prepared by grit blasting, the first coat of reinforcement primer should be applied within 3 hours and it should totally encapsulate the exposed reinforcement including locations where bars cross (Kay 1992). A good surface patching is shown in Figure 4.10(a), and a good quality of bonding between patched and old concrete is shown in Figure 4.10(b) by a core taken out of the patch. Surface preparation. The next step is to prime the surface of the concrete in the repair area. If cementitious repair mortars are used, the substrate should be thoroughly soaked with clean water before applying a bonding aid. When the bonding aid is applied, the surface should be damp but free of standing water. When polymer mortars are used as repair material, the concrete surface must be soaked with water. After that the first layer of repair mortar may be applied to the concrete surface. For cementitious mortars, the mortars must be applied when the bonding aid is wet. For polymer mortars, normally the bonding aid should reach a tacky condition before the application of repair mortars. The repair mortars have to be painted by hand, layer by layer, so that the repair mortars are in good compaction and uniform density. The thickness of individual repair mortar layer depends on the repair material as well as the orientation of the surface being repaired, typically 25–50 mm for vertical repairs and 20–30 mm for overhead repairs with normal weight mortars (Kay 1992). In order to obtain a good bond to the next layer, the surface of intermediate layers is usually left in a rough condition. As a general rule, the following layer is painted when the previous layer has stiffened sufficiently to carry the applied weight but before final setting. In cases when the installation of the following layer is delayed, the surface of the previous layer should be scratched rough and damped with water and a bonding aid should be applied before the following layer is painted. It is recommended that the final layer should be at least 10 mm thick and finishing work is required to match the surrounding concrete. The finishing on flat areas can be achieved by a wood or steel trowel to give the required texture. After completion of trowelling, the repaired
Conventional repair and strengthening
195
Figure 4.10 (a) A good surface patching on a concrete slab; (b) a good quality of bonding between patched and old concrete.
3
areas must be either sprayed with a resin-based curing membrane, or covered with polyethylene sheets held down around the edges and kept in position for 4 days (Perkins 1986). Curing for the repaired concrete. The last step is curing the newly cast repair concrete. The curing time and condition depend on the properties of the repair materials. Cement-based repairs usually do require some
196
Conventional repair and strengthening initial protection during curing. When ordinary Portland cement is used, all traffic should be kept off the repaired areas for at least 7 days; this can be reduced to 2 days if rapid hardening Portland cement is used. In cold weather, these periods normally should be extended by about 50 percent (Perkins 1986). If emergency repairs have to be carried out, high alumina cement can be used (Perkins 1986). Sprayed-on curing membranes cannot be used between layers of a multilayer repair because they prevent bonding between layers. Since the curing of resin-based repair materials involves heat evolution, i.e. the faster the curing, the greater the rate of heat evolution. The build-up of high thermal stresses is possible for large-scale repair work, and in this case, the thermal stresses will be released when the repair material cools down. As stated before, spray-on curing membranes or sheets of polyethylene tapes are helpful in protecting resin-based repair materials from cold, heat and rain. For the similar reason, a large volume of resin-based materials should not be mixed and applied in one batch.
4.3.4 Repair of cracks in concrete There are usually some cracks on the surface of a concrete structural member, such as a concrete wall or a reinforced concrete beam. If the crack width is very small, the crack may not have to be repaired. However, when the crack width becomes large, it is very important to evaluate the effect of the crack on structural performance and durability of the structure, and repair the crack if necessary. Depending on the location and orientation of a crack in a structural member, a large crack can significantly reduce the load-carrying capacity of the structure. For example, if a crack significantly reduces the strength of anchor bolts embedded in concrete, it should also be repaired. In general, a professional structural engineer should be consulted for the structural evaluation. In addition to the assurance for load-carrying capacity, crack repair is also necessary for long-term durability concerns of the structure, such as corrosion protection and bond strength protection. Therefore, crack width is an important factor in making a decision as to whether the crack should be repaired and how to repair. As a general rule, cracks in reinforced concrete wider than those allowed by building codes or specifications should be repaired or sealed. A number of codes of practice specify maximum permissible crack widths for various service conditions. Tolerable crack widths according to varying exposure conditions specified by ACI 224R-90 are shown in Table 4.8 (Emmons 1994). Under loading, the crack width in concrete members depends on loading conditions. In the latest version of the ACI code (ACI 224.1R-07), the following formula was given for calculating the maximum crack width under bending:
Conventional repair and strengthening
197
Table 4.8 Exposure conditions and tolerable crack width, ACI 224R-90 Exposure condition
Tolerable crack width
Dry air, protective membrane Humidity, moist air, soil De-icing chemicals Sea water and sea water spray; wetting and drying Water-retaining structures
w=2
冪 冢冣
fs s β d 2c + Es 2
(in)
(mm)
0.016 0.012 0.007 0.006 0.004
0.41 0.30 0.18 0.15 0.10
2
(4.3)
where w = maximum crack width, in. (mm); fs = reinforcing steel stress, ksi (MPa); Es = reinforcing steel modulus of elasticity, ksi (MPa); β = ratio of distance between neutral and tension face to distance between neutral axis and centroid of reinforcing steel (taken as approximately 1.0 + 0.08dc); dc = thickness of cover from tension face to center of closest bar, in. (mm); and s = bar spacing, in. (mm). For structural members under direct tension, the following formula has been suggested by ACI 224.1R-07 to estimate the maximum crack width:
冪
w = 0.1f s3 dcA × 10−3
冪
w = 0.0605 f s3 dcA × 10−3
(in.-bl)
(SI)
(4.4)
(4.5)
According to an NCHRP Study (NCHRP Report 380) and various FHWA publications, the acceptable crack width from a corrosion and durability standpoint is between 0.004 in and 0.008 in (0.1 and 0.2 mm) (Xi et al. 2003). British Standard 8110 suggests that the maximum crack width of 0.3mm may be acceptable or aesthetically acceptable. ACI 318 (1995), Building Code Requirements for Structural Concrete, has limited the acceptable concrete crack width to 0.013 in (0.33 mm) for an exterior crack and 0.016 in (0.41 mm) for an interior crack.
198
Conventional repair and strengthening
A fundamental weakness of this approach lies in the fact that the crack width at the surface of the concrete will always be greater than the interior cracks, or greater than the width at the reinforcement, which may not always be the case. The difference will depend largely on the loading conditions and the thickness of cover. Furthermore, cracks may be considered unacceptable in the following two situations: 1
2
Aesthetically unacceptable. In this case, since crack widths are usually very small, care must be taken as inadequate crack repair techniques often result in poor appearance upon completion (as shown in Figure 4.11). Non-watertight for the structure. Some structures require watertightness, such as reinforced concrete water tanks. Small cracks on the surface of concrete result in accelerated water penetration into the concrete, which reduces the service life of the structure and more importantly leads to possible leakage of the structure.
There are many crack repair methods as listed in ICRI Guidelines and ACI documentations, such as epoxy injection, gravity filling, routing and sealing, near-surface reinforcing and pining, grouting, drilling and plugging as well as crack arrest. These methods may be grouped into three basic types (Newman 2001): 1 2
“Glue” the cracked concrete back together by epoxy injection or grouting. “Stitch” the cracked concrete with dowels.
Figure 4.11 Appearance of a repaired concrete by improper crack repair techniques.
Conventional repair and strengthening 3
199
Enlarge the crack first and then “caulk” it with a flexible or semi-rigid sealant.
The choice of repair methods for cracks should take into account several factors, such as the cause and extent of cracking, the present condition of the crack, the location and environment of the cracks, and whether the crack is still actively moving or not. Typical examples of live cracks are those in movement joints. For example, cracks under wet-dry, industrial and marine conditions will require materials and methods quite different from those found acceptable for cosmetic repairs. Techniques that can be successfully used on horizontally cracked surfaces may not be effective on vertically cracked surfaces. For a crack induced by a one-time load application and which has stopped spreading can be repaired right away. For the cracks induced by shrinkage or settlement, the repair may be delayed so that further deformation can be minimized. In the case of live cracks, one must be aware that completely filling those cracks by injection will always lead to new cracking within the crack filler, on the interface with the cracked concrete, or within the old concrete (Bijen 2003). Repair of dead cracks wider than about 1 mm can usually be achieved by cement grouting although some traces may be left. Finer dead cracks are to be sealed by the injection technique (Allen et al. 1993). Inactive cracks can be repaired by epoxy injection or by sealing with mortar. For cracks which are deep or pass right through the member, crack injection can provide a satisfactory solution. 4.3.4.1 Resin injection Resin injection is used to repair concrete that is cracked or delaminated and to seal cracks in concrete to stop water leakage. Because of the high costs (generally about $200.00 per linear foot of injected crack), resin injection is not normally used to repair shallow, drying shrinkage, or pattern cracking. Two basic types of resin and injection techniques are used to repair cracked concrete. (A) EPOXY RESINS
Crack repair usually adopts the injection technique, because the resin used for the repair of cracks has high mechanical strength and resistance to most chemical environments encountered by concrete. Two methods of mixing and injection of resins for crack repair are commonly used. In the first method, the components are combined in a separate mixer and then transferred to the application equipment. In the second method, the components are pumped separately in carefully controlled proportions to a mixing head where mixing takes place automatically as the two components flow together. The latter method is suitable for large volume work in the situation
200
Conventional repair and strengthening
where continuity of the process over a long period is assured, while the former method is more suitable for a small repair job where the work is intermittent (Kay 1992). There are wide variations in the properties of different resins and it is therefore necessary to carefully choose to match the individual job requirements in relation to such matters as temperature of application, capability of bonding to moist concrete, shrinkage, thermal and elastic properties of the hardened resin and other special needs such as fire-resistance and high temperature stability (Campbell-Allen and Roper 1991; Trout 1997). ASTM C-881 provides more details on the types of resin that can be used for repair work. Also, the correct formulation of the resin is of vital importance and the requirements for the repair resin are likely to vary from job to job. The desirable qualities for repair resin include low viscosity, ability to bond to damp concrete, suitability for injection in as wide a temperature range as possible, low shrinkage, and toughness rather than high strength (toughness is the area under a stress–strain curve of the material) (Perkins 1986). Resin injection can be used to fill cracks and to bond together the concrete surfaces. The crack must be cleaned prior to the injection. Appropriate cleaning of the crack can be achieved by vacuum cleaning, or flushing of water or other solvents. When reinforcement rusting and concrete spalling have occurred, the best repair method may be removal of the defective concrete, cleaning of reinforcement, resin injection and then repair by mortars. The consistency of the materials used for crack repair must be adequate for the resin to penetrate into cracks and to provide durable adhesion to surfaces of cracks. Fluid resins are the most widely used repair materials for cracks. Important properties of any injection resin are its resistance to moisture penetration, to alkaline attack, a similar tensile strength to concrete and a similar modulus of elasticity with concrete. The tensile strength of the repair material should approach that of concrete as closely as possible. Before a resin injection can begin, the crack that appears on the surface of concrete member must be sealed, so that the liquid resin will not leak and flow out of the crack. During the injection process, epoxy resin is injected into cracks and micro-fissures in concrete and into voids between concrete and reinforcing steel. The bond between the epoxy and the concrete should be stronger than the concrete itself, so the repaired concrete should be as good as new concrete. It has been found that the injection of a low-viscosity epoxy is a possible repair method for cracks between about 0.02 mm and 6 mm in width (Warner 1977). For a crack width smaller than 0.02 mm, the epoxy cannot flow into the cracks. For small cracks, compressed air and solvents can be used for the cleaning work and for removal of water in the crack. (B) POLYURETHANE RESINS
Polyurethane resins are used to seal and eliminate or reduce water leakage from concrete cracks and joints. They can also be injected into cracks that
Conventional repair and strengthening
201
experience some small degree of movement. Such systems, with the exception of the two-part solid polyurethanes, have relatively low strength and should not be used to structurally re-bond cracked concrete. Cracks to be injected with polyurethane resin should not be less than 0.005 inch in width. Polyurethane resins are available with substantial variation in their physical properties. Some of the polyurethanes cure into flexible foams. Other polyurethane systems cure to semi-flexible, high density solids that can be used to re-bond concrete cracks subject to movement. Most of the foaming polyurethane resins require some form of water to initiate the curing reaction and are, thus, a natural choice for use in repairing concrete exposed to water or in wet environments. Polyurethane resin used for crack injection should be a two-part system composed of 100 percent polyurethane resin as one part and water as the second part. With appropriate water to resin mixing ratios, the resulting cured resin foam can attain at least 20 psi tensile strength with a bond to concrete of at least 20 psi and a minimum elongation at tensile failure of 400 percent. The manufacturer’s certification that the product meets these minimum requirements should be required before the injection resins are accepted for use on the job (ADWR 2006). (C) INJECTION EQUIPMENT
Resins can be injected using several types of equipment. Small repair jobs employing epoxy resin can use any system that will successfully deposit the epoxy in the required zones. Such systems could use a pre-batch arrangement in which the two components of the epoxy are batched together prior to initiating the injection phase with equipment such as small paint pressure pots. Large epoxy injection jobs and polyurethane resins injection jobs generally require an injection pump in which the two epoxy components are pumped independently of one another from the reservoir into the mixing nozzle. At the mixing nozzle, located adjacent to the crack being repaired, the two epoxy components are brought together for mixing and injecting. The epoxy used in this injection technique must have a low initial viscosity and a closely controlled set time. Several companies have proprietary epoxy injection systems, which allow them to make satisfactory repairs under the most adverse conditions. Contact information of these companies can be obtained from the Materials Engineering and Research Laboratory, Code D-8180, Denver, Colorado. (D) APPLICATION
Sealing is commonly achieved by providing a surface dam which completely bridges the crack. Resin injection requires a high degree of skill for satisfactory execution and is therefore normally carried out by specialist
202
Conventional repair and strengthening
contractors. As a general principle in sealing cracks by injection, it is suggested to start at one end and work progressively along the crack. A series of injection points are formed at intervals along the length of the crack and grout is injected into each point in turn until it starts to flow out of the next one. The point in use is then sealed off and the injection equipment can be moved to the next point, and so on until the full length of the cracks has been treated. Complete filling is assured when resin appears at all the injection points and vents. Sometimes, re-injection is necessary, because the injected resin may flow from the main crack into fine capillaries. This is especially true for high-pressure injection. Re-injection has to be accomplished prior to hardening of the previously injected resin. For large structural elements or structural elements with deep cracks through the structural thickness, both sides of the elements should be injected by resins in order to successfully repair the cracks. (E) QUALITY CONTROL AND EVALUATION
As a quality control procedure, concrete samples can be cored to assess the effectiveness of the repair in achieving desired objectives (see Figure 4.12). Coring concrete samples should be done through the crack plane. The concrete samples are indicators for evaluating the depth and completeness of resin penetration into the crack. The injection quality can also be checked by
Figure 4.12 A concrete core is taken from a bridge deck.
Conventional repair and strengthening
203
ultrasonic method. To conduct an ultrasonic test, one transducer can be placed at one side of a crack, and the other transducer on the other side, as shown in Figure 4.13. To evaluate the repair effectiveness, ultrasonic signal travel time between the locations of the two transducers can be measured prior to the repair, which provide baseline data and can be compared to those obtained after the injection. The baseline data of the signal travel time obtained prior to the repair should be larger than those obtain after the repair, because the repair work is supposed to heal the crack and the concrete between the two locations should become one solid piece after the repair (the ultrasonic signal travels faster in a solid without any cracks and voids). More than one location along the crack length should be tested for the signal travel time. The test data will reflect the average quality of the repair work. 4.3.4.2 Stitching Instead of gluing a crack together, repair can be done by stitching a crack together as shown in Figure 4.14, in which the crack is stitched by iron or steel dogs. Stitching can be considered as one of the crack arrest techniques. Other crack arrest techniques are described elsewhere, such as ACI 224.1R-07. When stitching a crack, a series of stitches are sufficient to make the total tensile strength of the repaired concrete equal to or greater than the
Figure 4.13 Ultrasonic testing of concrete cracking before repair.
204
Conventional repair and strengthening
Figure 4.14 Repair of a crack by stitching of concrete.
tensile strength of the original concrete. The best method of stitching is to bend bars into the shape of a broad flat-bottomed letter U between a foot and three feet long and with ends about 6 inches long or less if the thickness of the concrete is less. Before stitching, clean and seal the crack, and use a flexible seal for active cracks. Drill holes on both sides of the crack. The holes should not be in a single plane and the spacing should be reduced near the ends of the crack (because of the stress concentration at the crack tip). Clean the holes and anchor the legs of the dogs in the holes with a nonshrink grout or an epoxy. The stitching dogs should vary in length and orientation to prevent transmitting the tensile forces to a single plane. The following considerations should be made when using stitching: Stitch both sides of the concrete section where possible to prevent bending of the stitching dogs. Bending members may only require stitching on the tension side of the member. Members in axial tension must have the stitching placed symmetrically. Stitching does not close the crack, but can prevent its propagation. Stitching that may be placed in compression must be stiffened and/or strengthened to carry the compressive force, such as encasement of the stitching dogs in a concrete overlay. As an alternative to stitching at right angles sometimes it is necessary to use diagonal stitching since stitching at right angles does not resist shear along the crack. One set of dogs can be placed on each side of the concrete if necessary (Champion 1961). 4.3.5 Repair of concrete damaged by fire 4.3.5.1 Fire damage in reinforced concrete structures Generally, concrete is recognized as one of the most excellent thermalresistant building materials. Compared with other construction materials, concrete suffers less from a fire disaster, therefore concrete has been used in buildings and industrial facilities, nuclear reactor containment structures, and nuclear waste containers as a fire-resistant material. To determine the required concrete thickness for fire resistance and to determine if cracking or spalling is likely to occur in the situation of fire exposure, an understanding of the high-temperature behavior of concrete is necessary for designers. Fire usually causes damage to a large area of concrete on beams, columns and soffits of slabs. In the case of floor slabs and beams, it is the soffit which is normally the most affected. For reinforced concrete structures, the reinforcement may also be affected. For instance, it is often found that, after
Conventional repair and strengthening
205
fire, some of the reinforcements in the slabs buckle due to the relative smaller depth of the cover. This reinforcement must be replaced. The effect of fire on concrete depends on the temperature reached, the exposure period in fire and the characteristics of concrete itself in terms of cement type, water–cement ratio, cement content, aggregate type, and the thickness of concrete cover to the steel reinforcement. The structural damage of concrete structure caused by fire may be (1) spalling; (2) strength reduction in concrete and steel; (3) loss of anchorage of reinforcement; (4) excessive deflection of beams and slabs; and (5) distortion of the whole structural framing. The latter three types of damage may be so severe that demolition and replacement are sometimes the only possible solution (Campbell-Allen and Roper 1991). Generally, the strength of concrete is significantly affected by extreme temperatures occurred during a fire. Figure 4.15 shows the temperaturestrength curves of concrete. In Figure 4.15, all test data were residual strengths of concrete measured after the exposure to different high temperatures (Lee et al. 2008). The test data were then divided by the strength of concrete under room temperature, so they are relative residual strengths. The three curves represent the residual strength of concrete cooled by different methods after the high temperature exposures. One can see that water cooling resulting in the highest reduction in the relative residual strength in all temperature ranges except 800 °C, which implies that if a fire in a structure was extinguished by using large amount of water, the damage in the concrete is higher than that in a structure under fire and then cooled down naturally. The same phenomena can be observed in Figure 4.16 for the reduction of stiffness of concrete exposed to different high temperatures (Lee et al. 2008). It is important to emphasize that using water to extinguish fire is a commonly used method, and we do not mean this method should
Figure 4.15 Relative ultimate residual strength vs. maximum temperature.
206
Conventional repair and strengthening
Figure 4.16 Relative initial tangent modulus vs. maximum temperature.
not be used. One should realize the consequence of water cooling, which leads to high thermal gradient in concrete members and thus high thermal stress. Figure 4.17 shows the stress–strain curves of concrete exposed to different high temperatures, which indicates the deterioration of concrete in the full range of strains, including ascending and descending portion of the curve. The degradation of concrete under fire depends heavily on the type of aggregate used in the concrete. Heating rate is another important factor to consider when evaluating fire
Figure 4.17 Stress–strain curves of concrete exposed to high temperatures (natural cooling).
Conventional repair and strengthening
207
Figure 4.18 Concrete specimens exposed to maximum temperature 800 °C and cooled down by natural cooling.
damage. Figure 4.18 shows two damaged concrete specimens by fire with the same maximum temperature 800 °C and the same natural cooling method. One can see clearly that the one with slow heating rate of 2 °C/min. exhibits less surface cracks and smaller crack width than the other one with the heating rate of 15 °C/min. Under high heating rate, spalling damage of concrete occurs in the temperature range below 350–450 °C. Some severe fire disasters showed that progressive spalling of concrete may occur during a fire, when the concrete spalling takes place in the surface layer first and progresses into deeper part of concrete. As a result, the load carrying capacity of the structure can be severely reduced. High temperature results in significant damage to concrete as well as to steel. Therefore, the repair for fire damage of concrete structures may necessitate the addition of new reinforcing steel and link the new steel to the existing steel. It is important that a realistic assessment of the damage is made as soon as possible after the fire, since an immediate inspection will make it easier to evaluate the maximum temperatures reached in different parts of a concrete structure during a fire and the temperature gradient through the structural members. Maximum temperature reached in the structure can be estimated reasonably well by a careful examination of the debris after the fire. This can be compared with the temperature distribution and strength loss determined from test evidence. The detailed quantitative investigation using various tests is to determine the temperatures to which the member has been subjected and the loss of strength of the concrete as well as the reinforcement. There are three main tests which could be carried out in order to determine strength loss in concrete: (a) color determination; (b) Schmidt hammer; and (c) core testing. Change in the color of concrete resulting from heating
208
Conventional repair and strengthening
may be used to indicate the maximum temperature attained and the equivalent fire duration. In many cases at above 300 °C a pink discoloration may be readily observed. Pink or red spots are often observed on concrete exposed to a maximum temperature of 400 °C and above. Depending on the cooling method, the color of concrete subjected to slow and natural cooling is generally pink or red, while the color of concrete subjected to water cooling was dark pink or dark red. The onset of noticeable pink discoloration is important since it coincides approximately with the onset of significant loss of strength due to heating. Therefore, any pink discolored concrete should be regarded as being suspect. There are various books and industrial codes for performance and requirements of reinforced concrete structures under high temperatures (Bazant and Kaplan 1996; Phan 2004; Felicetti et al. 2004), such as ACI 216.1-07/TMS-0216-07, Code Requirements for Determining Fire Resistance of Concrete and Masonry Construction Assemblies; and ACI 349.1R-07, Reinforced Concrete Design for Thermal Effects on Nuclear Power Plant Structures. Figures 4.19, 4.20, and 4.21 depict the strengths of various types of concretes and steels under high temperatures specified by ACI 216.1-07. Table 4.9 shows the temperature dependence of hot-rolled reinforcing steel specified in Eurocode3 (ENV 1992-1-2, 1995a; ENV 1993-1-2, 1995b). The stiffness and strength values have been normalized by the respective values at reference temperature (i.e. room temperature).
Figure 4.19 Compressive strength of siliceous aggregate concrete at high temperature and after cooling (ACI 216.1-07).
Conventional repair and strengthening
209
Figure 4.20 Compressive strength of carbonate aggregate concrete at high temperature and after cooling (ACI 216.1-07).
Figure 4.21 Strength of flexural reinforcement steel bar and strand at high temperatures (ACI 216.1-07).
4.3.5.2 Repair methods of fire-damaged concrete Repair of fire-damaged reinforced concrete structures requires restoration of any loss in strength, durability, and fire resistance of concrete and steel. In situations where there is still sufficient strength and durability left for the
Conventional repair and strengthening
210
Table 4.9 Hot rolled reinforcing steel (0.2%), ENV 1992-1-2 (1995a) Temperature (°C)
E(T)/E(20 °C)
σpr(T)/σ0,2(20 °C)
fy(T)/σ0,2(20 °C)
20 100 200 300 400 500 600 700 800 900 1000 1100 1200
1.00 1.00 0.90 0.80 0.70 0.60 0.31 0.13 0.09 0.07 0.04 0.02 0.00
1.00 1.00 0.81 0.61 0.42 0.36 0.18 0.07 0.05 0.04 0.02 0.01 0.00
1.00 1.00 1.00 1.00 1.00 0.78 0.47 0.23 0.11 0.06 0.04 0.02 0.00
materials, a thin hand- or spray-applied repair material could be used to restore the structure. The commonly used repair materials for a firedamaged concrete structures are cementitious mortars, epoxy resinsmodified mortar or concrete. Caution should be used when epoxy-based repair materials is applied for the repair of fire-damaged concrete because the epoxy rapidly loses its strength if heated above 80–100 °C. Consequently, the most commonly used materials for the repair of fire damage are cementitious mortars and concrete. The repair of the fire-damaged concrete structure normally consists of the following stages: 1 2 3 4
Remove unsound concrete. Clean reinforcing steel and install new steel bars if needed. Clean and roughen the exposed concrete surface. Replace removed concrete.
The removal of unsound concrete and the roughening of the surface are usually carried out by hand tools and light pneumatic tools, or by sand blasting. The removal of concrete, cleaning and roughening of the new surface, and the cleaning of the reinforcing bars can also be done in one step by water jetting. After removing the cover concrete, the reinforcing steel is exposed. The rust on the corroded rebar can be removed by sand blasting device, hammers, or wire brushing. It is always a good idea to encapsulate the cleaned reinforcing steel in an alkaline coating prior to restoration of the concrete cover. If there is a significant loss of steel section, new steel bars should be used to replace the corroded bars. If the reinforcement steel has buckled after fire damage, it must be replaced. Three methods may be used for replacement of removed concrete: (a) recasting in formwork; (b) spraying or shotcrete; and (c) hand-applied
Conventional repair and strengthening
211
mortars. The choice of the repair method is usually determined by practical and cost considerations. Spraying and recasting are more suited for large volumes and large area applications, and when a speedy structural repair is required. However, if a high standard of surface finish is required, recasting is more preferred. The formwork must be well sealed against the existing structure, rigidly and firmly fixed. A high workability concrete should be used in order to have good placing and compacting of concrete. To replace concrete surfaces damaged by fire, shotcreting or guniting is the most frequently used technique. This method produces a dense concrete and no additional compaction is required after repair. Provided that the repairs are properly carried out, the shotcreting or guniting provides a high quality repair for fire-damaged concrete. The thickness of the sprayed concrete is largely dependent on the thickness of the original concrete cover needed for protecting reinforcing steel. It should be noted that the sprayed concrete may be applied over areas of undamaged concrete, and the surface of this must also be prepared by the methods aforementioned to be sound, rough and homogeneous. All traces of coatings must be removed since they would prevent a strong bond between the sprayed concrete and the existing concrete. It is always necessary that the sprayed concrete should incorporate a welded steel fabric, or steel or other fibres, to increase the bond between the existing concrete and the new concrete and to minimize the risk of cracks which allow the penetration of water into the concrete. Hand-applied mortars are more suitable for patch repairs and repair jobs of less volumes. Resin-based repair materials are normally applied by hand for small area repair. Before applying these mortars, the damaged concrete surface should be well treated by the methods described above. A thin layer of slurry grout coat of cement/latex can be applied first by brush to the prepared concrete surface. The repair mortars, normal cement-based or resin-based, are then placed on the coating layer by brush while the coating layer is still tacky. The repair should be built up in layers if the depth of the damaged concrete is large. The surface of the previous layer should be furrowed before the subsequent layer is applied. The final surface should be finished to match the surrounding surfaces. 4.3.6 Repair of corrosion-damaged reinforced concrete structures In industrial countries, the total cost of repairing metallic corrosion related deterioration is believed to be in the region of 4 percent of GNP. Therefore, how to repair corrosion-damaged concrete structures has become an important topic. For the onset of steel corrosion, there are four necessary conditions: enough moisture, air, low pH value, and enough ions (usually chloride ions from deicing salt) to form electrolyte. As one can see, moisture and air cannot be blocked from entering concrete. So, effort has been focused on
212
Conventional repair and strengthening
how to maintain high pH values and how to slow down the penetration of chloride ions into the concrete. High pH values in concrete mean high alkalinity in the concrete pore solution. Steel is passive under high alkalinity environment, and therefore corrosion will not occur. Then, there are two ways to trigger the onset of corrosion. One is by the penetration of carbon dioxide from the environment into concrete and the other is the penetration of water containing dissolved salts through the concrete cover or through a concrete crack. In the first case, the alkalinity of the concrete surrounding the steel could be reduced by the atmospheric carbon dioxide which reacts with calcium hydroxide in cement paste to form calcium carbonate (called carbonation of concrete). The reduction of calcium hydroxide leads to a low pH value. This creates an environment for the corrosion of steel to take place. In the second case, the alkalinity of concrete is not reduced, but when the chloride ion concentration is high enough, reaching a certain ratio with the hydroxyl ions (Cl/OH), the corrosion of steel may start. The required treatments for restoring the protective environment for steel depend on the extent and cause of the corrosion damage: 1
2
Carbonation-induced corrosion damage. Under such conditions, carbonated concrete should be removed and new concrete should be installed, re-passivation is provided by the new repair mortar or concrete. Chloride-induced corrosion damage. Under such conditions, if chloride has penetrated to the level beyond the steel reinforcements, removal of chloride around steel bars does not guarantee re-passivation as chloride ions may diffuse back from the deeper part of the concrete to the new concrete cover. This is the so-called redistribution of chloride after the repair. In this case, the repaired concrete will become cathodic and the rebar will be the anode. The corrosion will occur in the bars immediately. Other factors may influence the re-passivation of steel, for instance, coating of the steel reinforcements, and the application of membranes or sealers to limit the moisture content.
4.3.6.1 Restoration of concrete surrounding corroded bars The restoration of the corrosion protection system of the reinforcing steel can be accomplished by concrete, cement mortar or epoxy resin mortar. If the removal of concrete is selected, care must be taken to prevent the chloride redistribution. It is better to remove more concrete than needed to ensure all the chloride-contaminated concrete and all the carbonated concrete has been removed. Figure 4.22 shows the removal of the concrete under corroded steel bars by using hand tools. After the distressed concrete is removed, the condition of the rebars must be checked. For uncoated steel bars, the rust on the surface must be cleaned, and the fresh metal surface must be exposed, as shown in Figure 4.23 (a sensor is shown in Figure 4.23
Conventional repair and strengthening
213
Figure 4.22 Removal of the concrete under corroded steel bars using hand tools.
Figure 4.23 Steel bars after cleaning operation.
214
Conventional repair and strengthening
which was used to monitor the corrosion process). For epoxy coated bars, an inspection must be made to make sure the epoxy coating is not severely damaged. If there are damages and scratches as shown in Figure 4.24, the damaged areas on the coating must be repaired first by epoxy resins. Otherwise, the corrosion of steel bars will start from the damaged areas. The choice of the repair system depends on the thickness of concrete cover. For a repair with concrete, the use of sprayed concrete (shotcrete) is advisable. For restoration of the corrosion protection system with cement mortar, it is better to use polymer-modified cement mortars with high quality mix as the repair material, which can be placed in individual layers whose thickness does not exceed 5 mm. The water–cement ratio for the repair cement mortar is normally less than 0.4. If an adequate thickness of concrete cover cannot be achieved by placing the polymer modified cement mortar, an additional sealing of the concrete surface or a protective coating on the reinforcing steel will be necessary to prevent corrosion in the future. To make a high quality repair for corrosion protection, the water–cement ratio should be kept low. A low water–cement ratio will make concrete of reduced diffusivity. For Portland cement concrete, depth of carbonation decreases almost linearly with the water–cement ratio. The diffusion coefficient of chloride ions decreases with decreasing water–cement ratio of concrete or cement mortar. At the same water–cement ratio, the chloride ion diffusion coefficient decreases when cement content increases.
Figure 4.24 Damages and scratches on the epoxy coating of new steel bars.
Conventional repair and strengthening
215
The addition of pozzolanic materials such as fly ash and/or silica fume into concrete also helps improve the quality of repair concrete or cement mortar since these pozzolanic materials result in a low diffusion coefficient of the cementitious repair materials. Berke et al. (1988) have found that, at up to 10 percent addition by weight of cement with silica fume, the concrete exhibited significant improvements in properties directly affecting steel corrosion resistance. Compared with conventional concrete mixes, such concrete has much lower diffusion coefficient mainly due to the reduction of porosity. Sufficient curing is also of vital importance in improving quality of concrete and cement mortar, since insufficient curing will reduce durability against both carbonation and chloride ingress, and thus the ability of protection of reinforcing steel from corrosion. Epoxy resin mortars may be useful for small repairs with thin layer thicknesses. These mortars are able to prevent carbonation or the access of corrosive agents. Prior to restoration of the corrosion protection system with epoxy resin mortar, the concrete surface should be dry and the corroded reinforcing steel should be cleaned and should receive two coats of paint consisting of epoxy resin with active corrosion-inhibiting agents. 4.3.6.2 Cathodic protection methods Corrosion is an electrochemical process as discussed in Chapter 2. Anodic and cathodic areas develop on corroding materials or systems, with measurable potential differences between the anodes and the cathodes. Cathodic protection, an electrical system designed to stop corrosion by applying an electric current to the affected metal surfaces, has been extensively used to protect steel pipelines and tanks from corrosion for many years and has been applied in the protection of the reinforcing steel in concrete in recent years. The principle of the cathodic protection (CP) is to enforce the rebar to become a cathode and another metal as an anode. Since corrosion only occurs at the anode, the rebar will be protected. It should therefore be possible to reduce corrosion rates if the reinforcing steel under consideration can be shifted to a cathodic condition by some externally applied potential. This is the basic theory behind cathodic protection techniques. For practical applications, there are two basic CP systems: the surfacemounted anodes (or sacrificial anode) system and the impressed current system. (A) THE SACRIFICIAL ANODE SYSTEM
Sacrificial anode cathodic protection is a very effective system for many reinforced concrete structures where the concrete is often in a wet condition and electricity resistivity is low. In this method, the structure or the structural member to be protected is connected to a more reactive metal as shown in Figure 4.25.
216
Conventional repair and strengthening
Figure 4.25 A sacrificial anode cathodic protection system, called Galvashield cathodic protection system was installed in SH 85 SB in Greeley, Colorado.
Anodes of magnesium or zinc can be used in this way to provide protection to buried reinforcing steel. In this method, no additional power supply is needed, the small potential difference between the cathode and anode is sufficient to drive the electron and ion flow. The anodes corrode and are consumed in the process of providing protection and, for this reason, they are known as sacrificial anodes.
Conventional repair and strengthening
217
(B) THE IMPRESSED CURRENT SYSTEM
An external direct current supply can be installed as shown in Figure 4.26(a) for a concrete slab (such as a bridge deck). This type of CP system is called the impressed current CP system. The impressed current is DC with the positive side of the output connected to a purposely made anode, and the negative to the steel being protected. The anode can be of any material that can conduct electricity, but those materials that ensure system longevity are preferred, such as special metal alloys, those materials based on graphite, high silicon-iron, titanium, tantalum, niobium or the lead/silver alloys. Anodes in this type of CP system are comparatively inert and are designed to last much longer than sacrificial anodes. When correctly selected and used in a well-designed CP system, such anodes may have a useful life of 20 years and more (Perkins 1986). An additional power supply is necessary for this type of CP system exposed to the atmosphere where a higher potential difference is required. When this type of CP system is applied to reinforcement in structures above ground, to get effective cathodic protection, the reinforcement should be electrically continuous and the concrete between reinforcing steel and anode
(a)
(b) Figure 4.26 (a) Impressed current cathodic protection system installed in a concrete slab; (b) schematic for reinforced concrete cathodic protection.
218
Conventional repair and strengthening
should provide sufficient conductivity. The current flows through the concrete and the anode cannot be placed further away than the surface of the concrete. As a result, and due to the high electricity resistivity of concrete, it is necessary to use an anode system which is distributed across the surface if all of the reinforcement is to be protected. A schematic cathodic protection system for a reinforced concrete beam is illustrated in Figure 4.26(b). In the case of reinforced concrete superstructures, there is no appropriate electrolytic medium. The anode or anodes have to be distributed over the concrete surface if a reasonably uniform cathodic protection current is to be supplied to the reinforcement. Also, alkali-reactive aggregate should be avoided as cathodic protection may aggravate the reaction. The second type of CP system is more widely used than the first one. The procedures to install an embedded CP system include: 1 2 3 4 5
6 7 8
Assessment of the location of corrosion. Preparation of the concrete surface. Paint the rebar. Place the anode. Establish an electric connection between the reinforcement and DC power source and between anode and DC source. A control box as shown in Figure 4.27 is usually installed in a nearby location. Patch repair is applied. The DC source is activated. Verification tests are conducted to ensure proper operation of the CP system.
This type of cathodic protection system is generally used for the situation where reinforcement corrosion has happened because of the presence of high concentrations of chlorides in the concrete. If only cracked and spalled areas are repaired, a large area of reinforcement may remain in concrete with chloride concentrations sufficient to activate additional corrosion. Consequently, further cracking and spalling may occur. In such cases, the CP can only provide a limited means of prolonging the service life of the structure. If applied to the entire structure, the CP provides one of the most effective practical methods for reducing the corrosion rate to virtually zero. It is generally recognized that the CP system is a global approach to the corrosion problem and the only permanent repair of existing corroded reinforced concrete structures. The key to a successful cathodic protection of steel in concrete is to provide a uniform current density to the reinforcement. If the current density is too low, corrosion can occur; if it is too high, deterioration of the concrete around the reinforcement could occur. In cathodic protection, normally power is supplied to the anodes in the form of a very low direct current. For
Conventional repair and strengthening
219
Figure 4.27 The control box for an impressed current cathodic protection system.
new reinforced structures, cathodic protection is expensive compared to the use of better concrete. The CP method should be used with caution for pre-stressing steel because hydrogen could develop on the steel surface that is polarized as the cathode, which may cause hydrogen embrittlement of the high strength steel. Simultaneously, hydroxyl ions move away from the cathode and they may react with aggregate, leading to a reduction in concrete/steel bond. Therefore, cathodic protection is not recommended for pre-stressed concrete. 4.3.6.3 Realkalization techniques Realkalization is a technology to provide the long-term restoration of alkalinity to carbonated but otherwise sound concrete under the passage of an electric current. All areas of concrete damaged by corrosion must first be cut out and replaced. Realkalization is achieved by applying a potential across a temporary anode, external to the concrete, and an internal cathode – the reinforcing bar. The electrolyte, which is an alkaline solution of sodium or potassium carbonate, diffuses into the concrete towards the reinforcement. The electrochemical production of hydroxyl ions caused by the electric current creates more alkalinity at the surface of the reinforcement, which
220
Conventional repair and strengthening
repassivates the steel, and reduces the risk of further corrosion. The process normally takes between 3 and 15 days and it restores the alkalinity of concrete to a level greater than pH = 11.5, a level which supports passivity of the reinforcement. Figure 4.28 illustrates the process schematically using a shutter system to contain the electrolyte. There are several special considerations for this process, which will be described in Section 4.3.6.4 on chloride removal techniques. 4.3.6.4 Chloride removal techniques For chloride-induced corrosion damage, the chloride contamination can be removed from concrete as a repair process. The possible methods of chloride removal are: water treatment and chloride extraction (FIP 1991). Water treatment is based on the concept of chloride removal by water transportation. In this treatment, chloride contaminate is flushed out by a water stream as well as water dilution. However, since the permeability of concrete is very low compared to other porous media, it is difficult to remove the chloride content at greater depth. Chloride extraction is a method of corrosion control by extracting chlorides from around the reinforcement in otherwise sound concrete under the passage of an electric current. Chloride extraction, like realkalization, is achieved by applying a potential across a temporary anode, external to the concrete, and an internal cathode – the reinforcing bar. The positive anode attracts negatively charged chloride ions and the cathode repels the same. Chloride ions will either be removed from the concrete into the electrolyte or repositioned, away from reinforcement. In addition, the electrochemical production of hydroxyl ions creates more alkalinity at the surface of all the reinforcement, which repassivates the steel, and reduces the risk of further corrosion. Process run times are of the order of 3–15 weeks and electrolytes such as water or saturated calcium hydroxide are most often used. Figure 4.29 illustrates the process schematically using a shutter system to contain
Figure 4.28 Electrochemical realkalization.
Conventional repair and strengthening
221
Figure 4.29 Electrochemical chloride extraction.
the electrolyte. To avoid chlorine gas production at the anode, an ion exchange resin is mixed in the electrolyte (CaOH2). In this method, the concrete may be heated to 70 °C. Before applying realkalization or chloride extraction, any cracked, spalled or delaminated concrete needs to be cut out and replaced with material of similar electrical resistivity to the parent concrete. The exposed reinforcement should have been blast cleaned, but should not be primed. The processes involve the application of a temporary impressed current between anodes fixed to or near the surface of the concrete and the reinforcement. The anodes are normally of mild steel or activated titanium mesh. The anode is connected to the positive terminal of a direct current electric supply with the negative terminal connected to the reinforcement which thus becomes the cathode. Electrical low resistance continuity is important for both anode and cathode systems. The voltage used is normally between 10 and 40V. The electrolyte used is a solution of sodium or potassium carbonate for realkalization and water, sometimes with calcium hydroxide added, for chloride extraction. If alkali–aggregate reaction is a possibility, an alternative electrolyte such as lithium hydroxide can be used. The electrolytes are contained within an absorbent layer applied to the surface or within liquid tight shutters. The current flow, measured in amperes per surface area of concrete, is carefully controlled as is the total charge passed, both of which are related to the results produced. After completion of either process, the concrete surface may require cleaning. It is then beneficial to apply an anti-carbonation coating in order to maintain a stable environment and prevent further ingress of carbon dioxide or to coat and seal the concrete to prevent the ingress of chlorides. There are some side effects to be considered which include the following. Both processes increase alkalinity and may give rise to alkali–aggregate reaction if reactive aggregate is present. Where reactive aggregate is present,
222
Conventional repair and strengthening
proper consideration should be given and testing carried out before, during and after treatment in order to ensure that realkalization or chloride extraction does not accelerate the alkali–aggregate reaction. They both produce hydrogen at the reinforcement, which may cause hydrogen embrittlement of the steel if it is sensitive to hydrogen. The high alkalinity generated adjacent to the steel may temporarily reduce the bond between reinforcement and concrete. Some researchers have concluded that realkalized concrete is stronger, denser, less water absorbent and shows a tendency to lower levels of alkali silica expansion. Other research has indicated that any short-term loss of ductility of normal reinforcing steel would not cause significant problems and there is little or no effect on bond strength for deformed reinforcement. Realkalization and chloride extraction of pre-stressed and post-tensioned concrete are not normally recommended. 4.3.7 Repair of ASR affected concrete structures It is quite difficult to repair ASR damaged concrete structures. In general, blocking further moisture penetration into the concrete is the major mitigation method for repairing existing ASR damaged concrete. This is because research has shown that the gel formed during ASR is expansive only when there is a plenty of moisture involved in the reaction. Therefore, how to reduce the amount of water or moisture getting into concrete has become very important for ASR-affected structures. For the structures that are in contact with water or already have very high moisture content, it is extremely difficult to repair or to control the ASR-induced damage. Various sealants and waterproof membrane have been used to seal the surface of concrete. The problem with this type of method is that any small hole or damage to the membrane will defeat the purpose, because the amount of moisture needed for the formation of the expansive ASR gel is actually quite small. Depending on the extent of damage induced by ASR on a structure or a structural member, different methods can be used to repair the ASR-induced damage. For moderately damaged concrete structures, electrochemical methods can be used to mitigate ASR-induced damage and prolong the service life of the structure. For severely damaged concrete structures, the affected concrete must be replaced by new concrete with non-reactive aggregate and low alkali cement. More importantly, some additives should be used in the new concrete to reduce the risk of ASR damage. Otherwise, the new concrete will be affected very soon, for its working environment is the same as the old concrete. It is important to emphasize that if a concrete surface is affected by ASR, the interior concrete will eventually be affected by ASR, and it is just a matter of time before the entire structure is affected by ASR. Therefore the replacement of part of the existing concrete (the severely damaged part) by new concrete will not solve the problem, but only temporarily alleviate the ASR-induced damage.
Conventional repair and strengthening
223
4.3.7.1 Replacement of ASR-affected concrete Replacement techniques of ASR-affected concrete are similar to the techniques introduced earlier for repairing concrete surface. The difference is that special attention must be paid to the mix design of the new concrete to be used for the repair. Usually, there are two special requirements that should be satisfied. The first is that the cement to be used for the repair job should be low alkali cement, that is, the equivalent alkali content should be below 0.6 percent (ASTM C150). Low alkali cement is a necessary condition to reduce the risk of ASR in the new concrete. The second requirement is that the aggregates (sand and gravel) to be used in the new concrete should not be reactive. The reactivity of aggregates can be determined based on several ASTM standard methods, such as ASTM C1260, 1293, and 1567. In Canada, there are two standard methods for testing reactivity of aggregate, CSA A23.2-25A which is similar to ASTM C1260, and CSA A23.2-14A which is similar to ASTM C1293. In Japan, JASS 5N T-603 is a commonly used method for detecting reactive aggregates. In addition to low alkali cement and non-reactive aggregates, it is necessary to add some admixtures to reduce ASR potential. Pozzolanic admixtures, such as fly ash, ground granulated blast-furnace slag (GGBFS), and silica fume are often used. The function of added pozzolanic admixtures is to reduce the amount of calcium hydroxide in hardened concrete, and thus reduce the amount of reactants in ASR. Fiber reinforcement such as polypropylene fibers and steel fibers have been used to reduce the ASR-induced damage in concrete. The function of the fibers is to increase crack resistance of the concrete and thus the cracking time and crack width will be reduced in the ASR-affected concrete. Since the fibers do not alter the ASR damage mechanism, the expansion due to ASR will not be changed.
4.3.7.2 Lithium technologies Since the early 1950s, lithium compounds have been shown to be effective in mitigating ASR in concrete. Lithium compounds can be added to fresh concrete mixtures or hardened concrete, and the lithium ions do not stop ASR, but interfere with the expansive mechanism of the alkali–silica gel by changing the reaction product. Some of early experimental results (Durand and Gravel 1995) show that the use of lithium hydroxide (LiOH) affects the expansion due to ASR. Lithium combines with reactive silica to form a lithium-silica gel that does not absorb water and therefore does not expand. Hence, lithium compounds can be used for both the construction of new concrete structures and the mitigation of existing concrete structures affected by ASR (Folliard et al. 2003). Since 1995, there have been some projects in the US using lithium technologies to mitigate the ASR problem in pavement, bridge decks, and dams (Stokes 2002). The lithium ion, Li+, in several forms is capable of suppressing deleterious
224
Conventional repair and strengthening
ASR. Lithium hydroxide (LiOH), lithium fluoride (LiF), lithium carbonate (LiCO3), and lithium nitrite (LiNO3) have been used or studied for the mitigation of ASR expansion. LiNO3 is safe to handle, and has the least effect on concrete properties, while it has the greatest effect on concrete to prevent deleterious ASR. LiNO3 also does not generate hydroxide ions in concrete pore solutions, and therefore does not increase pH value of the pore solution. LiOH, on the other hand, is a hazardous material and difficult to handle. For an existing structure affected by ASR, a lithium compound can be sprayed on the surface of concrete. Lithium ions will penetrate into concrete and get involved in ASR. This method is effective only for very thin concrete members, because the penetration process of lithium ions is very slow and thus they can only reach a very shallow depth from the concrete surface. In order to accelerate the penetration process, an electrochemical treatment method similar to realkalization and chloride extraction may be used for mitigation of concrete slabs such as pavement and bridge decks (see the sketch in Figure 4.30). The technique involves the application of a high DC current density for a short period of time, typically a few days to weeks, between steel reinforcement acting as a cathode and an extended anode placed in an external electrolyte in contact with the surface of the concrete. Using this system, lithium ions can penetrate into concrete at a much faster rate than the spray application and become involved in the alkali–silica reaction. It should be pointed out that there is only a small window during the deterioration process of concrete due to ASR when the lithium technologies can be applied effectively. The window corresponds to an early stage of ASR damage in concrete. If severe ASR has already occurred in a structure, manifested by crack mapping on the surface of concrete, the concrete should be replaced instead of repaired by using lithium technologies. If the concrete structure is still in very good shape without any cracks, the penetration of lithium ions will be very slow even under the electrochemical system, and
Figure 4.30 The electrochemical repair system used in the highway industry.
Conventional repair and strengthening
225
thus the repair will not be effective. The best time (the window) for the application of lithium technologies is when the structure is diagnosed as at an early stage of ASR and there are only minor ASR-induced cracking. The cracking does not impose severe long-term damage in the concrete but allow accelerated penetration of lithium ions. 4.3.8 Repair of reinforced concrete bridge decks Bridge decks are usually made of reinforced concrete, while the girders and beams of a bridge may be made of steel, reinforced concrete, or pre-stressed concrete. In addition to environmental loadings such as freeze–thaw and wetting–drying cycles, bridge decks are exposed directly to traffic loading and attacks of de-icing salts in cold regions. Therefore, bridge decks are the structural members that need the most attention in terms of maintenance and repair in a bridge structure. Depending on the type of damage and the location of the damage, different repair methods can be used in practice. One important measure that should be taken to prolong service life of bridge decks is preventive maintenance or small-scale repair that can help significantly to improve the serviceability of bridges. 4.3.8.1 Preventive maintenance Preventing concrete deterioration is much easier and more economical than repairing deteriorated concrete. Preventing concrete deterioration begins in the design of the structure with the selection of the proper materials, mixture proportions, concrete placement, and curing procedures (Air Force 1994). Even a well-designed concrete structure will generally require follow-up maintenance actions. The primary types of maintenance for concrete are surface protection, joint restoration and cathodic protection of the reinforcing bars. Surface maintenance involves the application of sealers and coatings for protective purposes. It should be mentioned that coatings are usually different from sealers in the materials used, not in their thickness. Joint problems are usually treated with one of a variety of types of joint sealers, and cathodic protection was discussed in an earlier section, which involves the use of sacrificial anodes connected to the reinforcing bars or installation of an impressed current system. (A) SURFACE SEALERS
Surface sealers are applied to concrete for protection against chemical attacks by alkalines, salt solutions, or other chemicals. The actual need for a sealer must first be established, and then the cause and extent of any deterioration, rate of attack, and environmental factors must be considered when selecting the right sealer for the job. A variety of sealers are available for waterproofing and protecting concrete surfaces. Some of these products
226
Conventional repair and strengthening
have been successful in protecting new concrete from contamination by deicing salts and other harmful environmental agents. They have generally been unsuccessful at stopping the progression of already contaminated concrete (Utah DOT Research News 1998). (B) TWO TYPES OF SEALER AND THEIR APPLICATIONS
Linseed oil. A mixture of 50 percent linseed oil and 50 percent mineral spirits is normally used. The mixture is applied in two applications on a dry, clean concrete surface. The surface sealers should be less than 5 mils, and a test strip should be used to help determine the rate of application. The normal rate of application is about 40 square yards per gallon for the first application and 65 square yards per gallon for the second application. This treatment could last for 1 to 3 years under normal traffic condition. Silicone. Silicone has been used on concrete to minimize water penetration. Care must be taken where moisture has access to the back of the member and carries dissolved salts to the front face where it is trapped by the silicone. Silicone oxidizes rapidly and is somewhat water soluble. Treatments are required every 1 to 5 years. (C) APPLICATION PROCEDURES FOR SURFACE SEALERS
Specific surface preparation instructions provided by the manufacture of the selected sealer should be followed. The description in Section 4.3.2.1 should be studied. The report by NRC (National Research Council 1993, SHRP-S360) has a detailed description of the construction procedures related to the application of surface sealers. In general, a sealer should be applied to a clean and dry surface. The surface should be dry for maximum penetration of the sealers into concrete. Epoxy injection of visible cracks should be considered before the application of the sealer. An alternative is to first seal the cracks and then reseal the entire surface, including the cracks. To provide for proper penetration, the subsurface pore must be dry to the desired depth of penetration before sealing. This will help the penetration of sealers into a greater depth of concrete. The drying requires a sufficient period of dry, warm weather before application. Since the surface of concrete should be dry, water blasting is not recommended, and acid etching is discouraged. Newly patched and overlaid areas should be allowed to cure a minimum of 28 days after placing, or longer if recommended by the manufacturer. In any case, penetrating sealer should not be applied if the moisture content of the concrete is greater than 2.5 percent when tested in accordance with AASHTO T239 standard test procedure. Sealer may be applied by low-pressure pump and by flood and brush techniques. To ensure proper coverage, the area that needs to be sealed should be delineated on the deck surface. This area can be calculated from the capacity of the sprayer or container used for flooding the sealer. In
Conventional repair and strengthening
227
addition, the sealer should contain a fugitive dye to enable the solution to be visible on the treated surface for at least 4 hours after application. The fugitive dye shall not be conspicuous more than 7 days after application when exposed to direct sunlight. Sealer should not be applied when the temperature of the concrete surface is below 4 °C. Before sealing, exposed joint sealers and painted steel joints adjacent to areas to be sealed should be masked off. The sealed areas should be protected from rain and traffic spray for six hours after application. (D) SURFACE COATINGS
Plastic and elastomeric coatings form a strong, continuous film over the concrete surface. To be effective in protecting concrete, the coating must satisfy some basic requirements: the adhesive bond strength of the coating to the concrete must be at least equal to the tensile strength of the concrete; the abrasion resistance must prevent the coating from being removed; chemical reactions must not cause swelling, dissolving, cracking or embrittlement of the material; the coating should prevent the penetration of chemicals that will destroy the adhesion between the coating and concrete; for proper adhesion, the concrete must be free of loose dirt particles, oil, chemicals that prevent adhesion, surface water, and water vapor diffusing out of the concrete. 1
2
Epoxies. Epoxies are often used as coatings. As with most thin coatings and sealers, a protective overlay or cover is required if they are exposed to traffic wear or abrasive forces. Asphalt. Asphalt is used as a protective overlay for bridge decks. It provides water protection and a protective wearing surface.
(E) APPLICATION PROCEDURES FOR COATINGS
The surface preparation methods are similar to those suggested for surface sealers. For epoxy coating applications, the coating should be mixed to produce a uniform and homogeneous mix. For spray application, a lowpressure spray gun should be used. The coating should be applied when the temperature is between 50 and 90 °F. Two applications of the coating should be applied to ensure even coverage and minimize the likelihood of pinholes. Each application should produce a dry film thickness of 2–3 mil. The second coat is normally applied 24 hours after the application of the first coat, but this can vary with environmental conditions and material type. The coating should be applied at the rate of coverage approved during the material acceptance tests. For asphalt coating, the thickness of the coating is usually in the range of 2–4 inches. The application is similar to asphalt pavement overlay, which will not be described in detail here.
228
Conventional repair and strengthening
4.3.8.2 Repair methods for crack reinforcement Crack reinforcement is primarily used to repair cracks in the load-bearing portion of the structure for active and dormant cracks. This repair bonds the cracked surfaces together into one monolithic form. Repair procedure is as follows: 1
2 3
Clean and seal the existing crack with an elastic sealer applied to a thickness of 1/16 to 1/32 inch and extending at least ¾ inch on either side of the crack. Drill ¾-inch holes at 90 degrees to the crack plane, fill the hole and crack plane with epoxy pumped under low pressure (50 to 80 psi). Place a reinforcing bar (No. 4 or No. 5) into the drilled hole with at least an 18-inch development length on each side of the crack.
There are other repair methods for cracking in bridge decks, which are similar to those described in Section 4.3.4. Details can be found in ACI 224.1R-07. 4.3.8.3 Spall repair methods Spalling of concrete is a localized damage in concrete structures. Spalling of concrete cover outside of reinforcement bars is usually caused by corrosion of the steel bar. Another form of concrete spalling is due to high heating rate when the concrete is under fire. Patching methods are often used to repair the spalling damages. For bridge decks, the deteriorated areas may include the top layer of steel bars or both the top and bottom layers of steel bars. If only the top steel bars are corroding, a partial-depth repair may be used. For partial-depth deck repairs, the deteriorated concrete should be removed to the depth required to provide a minimum of 1.9 cm clearance below the top layer of steel bars. Maximum depth of removal for a partial-depth repair should not exceed half the deck thickness (National Research Council 1993, SHRP-H-356). Corrosion of both the top and bottom layers of steel bars requires full-depth repairs. In this case, the concrete within the delineated area for the entire deck thickness, normally 8 in. should be removed. It should be pointed out that patch repair of bridge decks has a relatively short service life because they do not address the cause of the problem, i.e. corrosion of the reinforcing steel, but address only the symptoms: spalling and delaminations. It is also important to note that when concrete contaminated with chloride beyond the threshold level is left in place in the area surrounding the patches, patches often accelerate the rate of deterioration of the surrounding concrete. The patched concrete area acts as a large non-corroding site adjacent to corroding sites and increases the rate of corrosion. Therefore, it is better to install a cathodic protection system as shown in Figure 4.25 for
Conventional repair and strengthening
229
new patches. The following are the construction procedures for partialdepth repair (FHWA 1999). (A) DELINEATION OF AREAS TO BE REMOVED
Areas of unsound concrete should be located using drag chains and hammer. The unsound concrete is evidenced by a hollow sound. The area to be removed should be delineated by the site engineer as the unsound or delaminated area plus a periphery of 3–6 in. The delineated area should be outlined with a saw cut 0.75 in. deep. Care should be taken to avoid cutting existing reinforcing steel. In no case are feathered edges acceptable. (B) REMOVAL OF UNSOUND CONCRETE
Many kinds of tools can be used to remove unsound concrete, which usually include: pneumatic breakers, milling machines, and hydrodemolition (National Research Council 1992, SHRP-S-336). These removal methods are for large scale bridge deck demolition, and they are different from those described in Section 4.3.2.1 for small-scale repair preparation of structural members in buildings. Each method has very specific strengths and weakness with regard to work characteristics, production, economics and quality. The following is a brief summary: • • •
Pneumatic breakers. They are the most expensive method but also the most flexible. They can be used for all sizes and shapes of area, to all depths and all bridge structural elements. Milling machines. These provide the least expensive method of concrete removal but they are also the most inflexible. They can only be used to remove large areas of surface and/or cover concrete on decks. Hydrodemolition. This technology lies between pneumatic breakers and milling machines in terms of cost and flexibility. Surface, cover, matrix and core concrete can be removed, but economies are only realized if work is done on large horizontal areas such as decks.
Quality, availability, flexibility, total coat and contractual risk can easily override the technical aspects. These vary with time and location and thus it is extremely difficult to make any firm rules as to which method must be used under general circumstances. (C) PREPARATION OF REPAIR CAVITY
Once all the unsound concrete has been removed, the cavity should be blasted clean to remove all loose material and provide a dust-free surface. All exposed reinforcing steel should be blasted to near white metal. With widely used epoxy coated rebars, one precaution should apply, that is, the rebar after sand blast should be re-coated with epoxy, otherwise, the
230
Conventional repair and strengthening
exposed steel due to the blast will be the new site for corrosion. Reinforcing bars with greater than 20 to 25 percent sectional loss as determined by the engineer should be lapped with reinforcing bar of equal diameter for 30 bar diameters on either side of the deteriorated area. Following the completion of blasting, the cavity should be air blasted to remove all dust and debris. The compressed air source used for blasting should have all oilers removed and oil traps installed. Full depth patches will require formwork constructed to prevent leakage of mortar. Patches should be reinforced with wire mesh attached either to reinforcing bars or dowels to secure the patch to the old concrete. Loose reinforcing bars should be tied at each intersection point to prevent relative movement of the bars and repaired concrete due to the action of traffic in adjacent lanes during the curing period. If new reinforcement is required, an adequate length to attain a lap splice (30 times the bar diameter) must extend from the existing section. If a proper splice is not possible, holes must be drilled into the existing concrete and dowels or anchors installed. (D) APPLICATION OF BONDING AGENTS
A good interface must be established between the existing and new concrete. For instance, the following options can be considered: 1 2
Epoxy bonding. Ensure the surface is clean, dry, and free of oil. Apply the epoxy agent to the prepared surface. Grout or slurry. Clean the prepared surface and saturate with water. Remove all free-standing water with a blast of compressed air, and apply a thin coat of grout.
(E) PLACEMENT AND CONSOLIDATION
The patch material may be batched and mixed at the site or supplied readymixed. The patch material should be placed in a manner that will prevent segregation, and consolidated with an internal vibrator. The surface of the patch should be floated and textured to match the surrounding concrete surface. (F) CURING
Portland cement concrete and hydraulic mortar/concrete must be moist cured for a minimum of 72 hours. Moist curing should be provided by clean, wet burlap covered with plastic sheeting anchored around the patch perimeter with sand. Shorter moist curing times may be approved by the site engineer in rapid patching conditions. For rapid repair materials, the patches should be cured until the structure is to be opened to traffic or for the minimum time recommended by the manufacturer.
Conventional repair and strengthening
231
4.3.9 Repair of reinforced concrete pavements There are many types of damage in concrete pavement (NCPP 2004). Figures 4.31(a) and (b) show the so-called “D” cracking and map cracking in concrete pavements. To extend the service life of concrete pavement, preventive maintenance is very important. The preventive maintenance methods used for concrete bridge decks such as sealers and coatings can also be used for concrete pavement. However, the cost required for large-scale application of the preventive maintenance methods will be quite high. Fortunately, the serviceability of concrete pavement can usually be kept very well for a long period of time without requiring any major repair. Pavement rehabilitation encompasses major and minor repair activities. The major repair activities are different from periodic maintenance activities. Major repairs will be viewed as any work that is undertaken to
Figure 4.31 (a) Severe “D” cracking in a concrete pavement; (b) map cracking in a concrete pavement. Source: (NCPP 2004).
232
Conventional repair and strengthening
significantly extend the service life of an existing pavement through the so-called 3R program: Restoration, Resurfacing, and Reconstruction. Restoration is used at an early stage when a pavement has little deterioration, and repair work is limited to isolated areas of distress. Resurfacing is used when the pavement has medium to high levels of distress and when restoration is no longer effective. Reconstruction is used when the pavement has high levels of distress and when resurfacing is no longer effective. Minor pavement repair activities include: thin asphaltic overlays, routine sealing of cracks and joints, slab sealing, pavement patching and pothole repairs. There are no definite equations, guides, or step-by-step procedures that one can use as a recipe for proper rehabilitation of concrete pavement. There are no “right” and “wrong” solutions to pavement rehabilitation problems, but rather “better” or “optimum” solutions. Optimum solutions maximize benefits while minimizing costs, which is often not attainable due to the constraints imposed. A realistic procedure is as follows: 1 2 3
Determine cause of the distresses or pavement problems. Develop a candidate list of solutions that will properly address the problem. Select the preferred rehabilitation method given economic and other project constraints.
Reconstruction involves concrete pavement design (rigid pavement design), and resurfacing involves thickness design for overlay, which is based on concrete pavement design. Both of them are beyond the scope of this book. We will only briefly introduce resurfacing methods and then discuss in detail a restoration method in this section. 4.3.9.1 Concrete overlay Based on traffic loading, the thickness of concrete overlay can be calculated. Then the construction method of a concrete overlay is similar to the construction method for a concrete pavement, in which the old concrete pavement is treated as a sub-base. Depending on the material used for the overlay, it is called white topping if the overlay is made of Portland cement concrete and black topping if it is made of asphalt concrete. Many attempts have been made to reduce the permeability of concrete overlay on bridge decks and concrete pavement so that the penetration of chloride ions into the concrete can be slowed down. Low-slump dense concrete overlay has been used by many Department of Transportations (DOTs), in Iowa, Kansas, Minnesota, and New York. High performance concrete overlays with silica fume, microsilica, fly ash, slag, latex have been used by many DOTs. Where a bare concrete deck is desired, Region 6 of Colorado DOT has been topping the deck with 2 inches of silica fume
Conventional repair and strengthening
233
concrete. Recently, very early strength (VES) latex-modified concrete overlay has been used by Virginia DOT. Various fiber reinforced concretes have been used by Oregon and South Dakota DOTs for concrete overlays. In Europe, ultra-high performance fiber reinforced concrete (UHPFRC) has been used for in the overlay of concrete decks, which has very low permeability and very high tensile strength. High tensile strength helps to reduce cracking tendency of concrete. However, the initial cost of UHPFRC is quite high because of the use of large amount of fibers and fiber dispersion additives in UHPFRC. Polymer concrete has also been used for overlays. They are mixtures of epoxies with aggregates such as polyesters with aggregates, methacrylates with aggregates, and polyurethanes with aggregates. Similar to UHPFRC, the initial cost of polymer concretes are relatively high. 4.3.9.2 Full-depth repair – a restoration method Restoration methods are also referred to as non-overlay pavement rehabilitation methods, which include the following: • • • • • • • • •
full-depth repair; partial depth patching; joint crack sealing; subsealing – undersealing; grinding and milling; sub-drainage repair; pressure relief joints; load transfer restoration; surface treatments.
Some of above listed methods are similar to those described in the last section for bridge deck repairs, or similar to Section 4.3.3 and Section 4.3.4. They will not be repeated here. We will only introduce in detail the full-depth repair. FULL-DEPTH REPAIR
The full-depth repair consists of five basic steps: 1
Locating and isolating the patch area. Good judgement is essential in defining the limits for full-depth repairs. Each repair should be large enough to replace all significant distress, resist rocking under traffic, and be easy to work in, and small enough to minimize the material cost. Typically, a patch that is of full-lane width and half-a-lane width long meets these criteria. For utility cuts, the patch length and width should be slightly larger than the planned trench length and width. This creates
234 Conventional repair and strengthening
2
3
4
5
a shoulder around the utility cut that keeps the excavation from undermining the existing concrete during utility repair and helps support the concrete patch after placement. Removing the damaged concrete. There are two procedures to remove the old concrete: the lift-out method and the break-up and clean-out method. With the lift-out method, the old concrete is lifted out in large sections. With the break-up and clean-out method, the concrete is broken into small pieces and removed. The lift-out method generally is faster and causes less sub-base disturbance. The break-up and clean-out method removes severely damaged concrete that the lift-out method cannot remove. However, this method may damage the sub-base, and so may require more sub-base repair than the lift-out method. Preparing the patch area. With the lift-out method, the base repair generally consists of shovelling up the damaged materials from the opened area. With the break-up and clean-out method, base repair includes recompacting the sub-grade and adding granular sub-base layers. If the area to be repaired includes a joint, tie bars or dowel bars should be used. Tie bars are deformed steel bars grouted into the existing concrete, while dowel bars are smooth steel bars inserted into holes drilled in the existing concrete. Both tie bars and dowel bars are installed for transferring the load between the joints. For a pavement with little truck load, the load transfer can be done by aggregate interlocking, which is the action between the roughened face of the old concrete and the face of the patch. The roughened face can be created by chips along the patch edges of the old concrete. The sub-base should be excavated out from beneath the concrete slab and replaced with new concrete. Placing and finishing the new concrete. Generally, the patch thickness is the same as the existing slab thickness. Normal Portland cement concrete and high early strength concrete have all been used for full-depth repair. The strength of the new concrete should not be lower than the old concrete. Asphalt concrete is not a good material because it does not last as long as Portland cement concrete. For continuously reinforced concrete pavement, there is no joint along the slab. Usually the repair work should be done in the afternoon in the summer season to avoid crushing failure under large thermal expansion. Proper curing of the patch is important. It is best to begin curing operations as soon as possible after completing the finishing operation. Typically, either a clear or white-pigmented curing compound may be used. For high early strength patches, insulation mats should be placed over the repair during curing to increase the strength gain of the concrete. For utility cut, flowable fill is an ideal alternative to base repair. Flowable fill is a low-strength, selflevelling material made with cement, fly ash, and sand that flows in and around repairs and then hardens. Since it is designed not to become too hard, it is easy to remove later. Opening to traffic. The proper time to open a patch to traffic depends on
Conventional repair and strengthening
235
the strength gain of the patch material. Generally, full-depth repairs can be opened when the strength of new concrete has reached at least 20 MPa (3000 psi), or the strength has reached the design value specified by the engineer. Most of concrete mixtures achieve this strength in 24–72 hours.
4.4 Strengthening techniques Strengthening of a concrete structure may be required due to several reasons: 1 2
3
Change of usage which may cause over-stress in the structural member. Serious materials and structural deteriorations which cause structural members to be no longer able to carry the imposed loads with an adequate factor of safety. A combination of (1) and (2).
Strengthening of structural members can be achieved by replacing poor quality or defective material with better quality material, by attaching additional load-bearing material, such as high quality concrete, additional steel, thin steel plates, various types of fiber reinforced polymer sheets, and so on, and by the redistribution of the load such as by adding a steel supporting system. The purpose of strengthening is to increase the load-carrying capacity or stability of a structure with respect to its previous condition. 4.4.1 Design considerations for strengthening A repair or strengthening work starts with a diagnosis, or evaluation of the structural condition, and then selection of materials and the strengthening method, preparation of the areas in the structure to be strengthened, and application of the repair or strengthening. It is necessary to conduct a careful investigation and assessment of the condition of the existing structure before any strengthening work. It is usually essential to check carefully the position, condition and amount of the reinforcement with the help of original drawings and design data. If the original documentations are not available, some non-destructive testing methods may be used, as described in previous chapters. The information obtained from the structural assessment is necessary and useful for the design of strengthening. In general, the strengthening of structures should be designed and constructed in accordance with appropriate design codes. In design of strengthening work, typical problems that should be solved are the transfer of shear forces between the old concrete and the new concrete applied for strengthening reinforcement, and the post-tensioning of the existing structure which in some respects is different from the post-tensioning of a new structure. Interactions between new and old concrete should be always considered
236
Conventional repair and strengthening
in a design for strengthening of concrete structures or structural members. As a general rule, the aim is to get structural parts, composed of different concretes, old and new, to act as a homogeneously cast structural component after strengthening. To achieve this, the interface between the old concrete and the new concrete must be capable of transferring shear stresses without relative movement of such a magnitude that the structural performance is significantly affected. In addition, the joint must be durable for the severe environment. In strengthening work using a large volume of concrete, additional stress may be produced due to the heat of hydration in new concrete. Temperature differences between old and new concrete can be limited by special measures, such as by pre-heating the old structural elements and/or by cooling the fresh concrete. Differences in shrinkage and creep between old and new structural elements require careful evaluation. Suitable mortars or concrete with low shrinkage and creep properties and minimal development of the heat of hydration should be employed for strengthening. An effort should also be made to match, as closely as possible, the strength and modulus of elasticity for the two concretes. Of course, compatibilities of the old and the new concretes should also be seriously taken into consideration when strengthening a job with large volume of concrete. Strengthening of reinforcement subjected to tensile forces can be achieved by several methods, which will be discussed in detail in the following sections: 1 2 3
Additional reinforcement placed in the old cross-section or in an additional layer. Replacement of corrosion damaged rebars. Addition of epoxy-bonded steel plates.
4.4.2 Addition of reinforcing steel In the simple case, a strengthening of the concrete tension zone is possible by means of addition of rebars. Rebars should be added after unloading and after the cover has been removed. Sometimes, additional support is needed for the structures and/or structural members when a concrete structure is strengthened by adding reinforcement. After additional rebars have been placed, the concrete cover must be re-established. Effective anchoring for rebars is required. Welding is recommended when the space between the rebars is limited. For a welding joint, attention should be paid to the heat propagation in the bar that could create serious problems in old concrete. The uneven elongation of the heated bar and concrete may lead to high local tensile stress in concrete that could result in cracking and debonding of the rebar. It is therefore suggested that the anchorage length of the bar should be increased by six bar diameters.
Conventional repair and strengthening
237
Strengthening of the reinforcement by incorporation of additional steel has mainly the following disadvantages: (a) it is a labor-intensive and timeconsuming job; (b) it needs significant space for the retrofitting process to be performed; (c) it may affect the operation of the existing facilities; and (d) it may reduce space or headroom after retrofitting. The following preparation work is important for the addition of reinforcing steel: 1
2
3
4
5
Concrete around corroded bar should be removed to a 50 mm distance beyond the corroded portion. The reinforcement steel bar should be exposed with at least 20 mm clearance below. If concrete around non-corroded bars are damaged to an extent that will affect concrete/steel bond, the affected steel should be completely exposed. Rust should be removed from corroded steel surface. Grit blasting and needle gun are effective but they will produce air and noise pollution. A high pressure water jet can be used while wire brushes are not so efficient. After rust if removed, dust on steel surface can be removed with an industrial vacuum cleaner. If there is a concern about insufficient compaction of repair material, which may lead to incomplete contact of repair material with steel and ineffective re-passivation, the steel surface can be treated with alkaline slurry first. Care should be taken as slurry may thicken quickly and not bond well to the repair material. Polymer cement slurries can be used, but they may reduce alkalinity or insulate steel from the passivating repair mortar or concrete. For reinforcement steel in concrete structure, when re-passivation is not possible, the electrical isolation of steel from the surrounding materials may be helpful. However, the preferable technique for treatment of reinforcement steel is the use of zinc coating to provide cathodic sacrificial protection. Epoxy coating can also be employed, but there may be the following problems: (a) the steel surface may not be completely covered in the field; (b) corrosion develops in the uncoated part which becomes an anode; and (c) penetration of water into steel/epoxy interface may lead to debonding and corrosion.
4.4.3 Replacement of the reinforcement If reinforcement steel inside the old concrete structure has been severely damaged, or has lost a substantial proportion of its cross-sectional area through corrosion, it may need to be replaced by new rebars. Pullar-Strecker (1987) suggests that replacement of steel is necessary if it has lost more than 20 percent of its cross-section area but other researchers require replacement if more than 10 percent of the cross-section area is lost (Campbell-Allen and Roper 1991).
238
Conventional repair and strengthening
Replacement of the reinforcement is more complicated than the addition of new rebars discussed in the previous section, because in this operation, the existing reinforcement must first be removed. Before any existing steel is removed, it is extremely important to check that all aspects of design have been considered before this step is taken. The structure should be unloaded, and additional support provided. In replacement of steel, it is necessary to open up sufficient concrete to provide adequate lap lengths with the undamaged portion of the steel bars by lapped splices, by welding or by coupling device. The replacement steel bar should be of the same quality as the original bar. If a different steel must be used, it should certainly not be so different that corrosion cells can be set up between the two adjoining bars. For instance, stainless steel bars should not be used to replace or reinforce conventional steel. Otherwise, the conventional steel next to the new stainless steel will immediately become an anode and the corrosion will start right away from the existing steel. If it is convenient to saw into sound concrete, new bars may be conveniently anchored in concrete placed in dovetailed slots. A longer anchorage length than that used in conventional design should be provided in all cases of replacement or addition of reinforcing material. The concrete cover should certainly be re-established after the damaged rebars have been replaced. In reinforced concrete beams or columns where the damage is confined to the steel stirrup, the new steel stirrup may be hooked round existing main corner reinforcement. In such situations, an extra thickness of concrete has to be anchored on while the structure may not allow an increase in the cross-section of the beam or column. A good solution may be to drill into the sound concrete and anchor the steel stirrup in epoxy. If this measure is adopted, the cause of reinforcement corrosion must be completely eliminated as a part of the repair procedure, since otherwise the new steel stirrup will be at risk of corrosion where it emerges from the epoxy anchorage (ibid.). When the new rebar just replaces the old deteriorated reinforcement steel, it carries no load until the structure is reloaded and the stress is transferred to it through the surrounding concrete or other anchorage. Since it will be difficult to determine how effective the transfer is going to be, generous allowances should be made for this uncertainty when assessing the final capacity of a structure repaired by replacement of reinforcement. 4.4.4 Epoxy-bonded steel plates The strengthening of concrete structures by means of epoxy-bonded steel plates has been used in many countries to provide additional load-carrying capacity or additional stiffness for the concrete structures, and to control flexural crack widths and deflections (McKenna and Erki 1994; Colotti and Spadea 2001). The advantages of this method can be attributed to the availability of high quality epoxy resins, and the minimum changes in geometric
Conventional repair and strengthening
239
dimensions of structural systems. The disadvantages of the method of strengthening by epoxy-bonded steel plates are: (a) it is a labor-intensive and time-consuming work; (b) it needs significant space for the retrofitting process to be performed; (c) it may affect the operation of existing facilities; and (d) it may reduce space or headroom after retrofitting. Also, additional measures are needed to prevent steel from corroding when applying this strengthening method. The gradual loss of strength of ambient cured resin adhesives at temperatures above 60 °C also limits the use of this strengthening method. If an epoxy-bonded steel plate is used for strengthening the soffits of beams and slabs, the steel plate could become detached from the concrete surface during fire due to the deterioration of adhesive. Under such situations, a secondary positive fixing system between the steel plate and the concrete surface should be provided. For instance, the steel plate can be bolted into concrete, providing mechanical anchors as the secondary fixing system. When employing this strengthening method, the following parameters and main performance criteria for epoxy resin adhesives are important (Perkins 1986): 1
2 3 4
The concrete surface must be well prepared to receive the epoxy and the surface of the steel plate must also be well cleaned and polished to a high standard using grit blasting. The epoxy must be carefully selected for both concrete and steel. The entire steel plate surface in contact with the concrete must be covered with epoxy. After the strengthening, if failure occurs, it will occur in the concrete because the shear strength of the cured epoxy is usually higher than that of the concrete.
After the concrete and the steel surface are cleaned, the installation of the steel plate is quite straightforward. First, the epoxy is applied within a short period of time on the surface of concrete, and the steel plate is then pressed against the epoxy-coated surface. The plate must be held in position until the epoxy has cured and gained enough strength. After the installation of the steel plate, the behavior of concrete structures strengthened by means of epoxy-bonded steel plates should be examined in two aspects: short-term behavior and long-term behavior. (A) SHORT-TERM BEHAVIOR OF STRUCTURES STRENGTHENED BY EPOXY-BONDED STEEL PLATES
The load-carrying capacity of the strengthened structures depends on the strength of the reinforcement, the old concrete, and the adhesive (epoxy). Yielding of the reinforcement will cause the adhesive to fail, but the utilization of high-strength reinforcement may not be helpful. As mentioned
240
Conventional repair and strengthening
above, the shear strength of the cured epoxy is likely higher than that of the old concrete, so that the strength of the old concrete has a large influence on the efficiency of the strengthening method because the failure is usually located within the concrete. To ensure a good bond between the epoxy and the old concrete, the surface of the old concrete should be treated to make sure that it is free from any defects likely to have a detrimental effect on the transfer of stress between concrete and steel plate. A common failure pattern of a strengthened structure is the slip between the reinforcing element and concrete. Under a one-dimensional tensile load, the slip will start at the loaded end of the reinforcing element and move, with increasing load, to the center of the element. Failure occurs suddenly, by abrupt elongation of the slip interface up to the end of the reinforcing element (peeling off of the plate). In correctly designed structures with bonded plate reinforcement, a ductile failure with yielding reinforcement can be attained. Theoretically, higher bond stress is to be expected from an increase in the elasticity of the reinforcing element and a decrease in the elasticity of the adhesive. Geometric influences are primarily the dimensions of the steel plate. The length of steel plate has a strong influence on the bond stress intensity, which decreases with increase in the length. The bond stress will also be influenced by the thickness of the plate. Therefore, glued-on reinforcing elements behave differently from the deformed bars, which can be designed using the same permissible bond stress for all diameters. With increasing width, there is a risk of defects in the adhesive and an increase of width results in a reduction of bond strength. Therefore, the width of the steel plate is normally limited to 200 mm (FIP 1991). The thickness of the adhesive coat, within a range of 0.5–5 mm, has no significant influence on ultimate load. The concrete dimensions, according to previous tests, do not appear to have any decisive effect on the ultimate load. The surface condition of the steel is an important parameter to achieve a successful repair. Suitable conditions can be achieved by sand-blasting. As a cleaned surface corrodes rapidly, a primer coating should be applied immediately which serves as a corrosion-protection layer and as an adhesive base for epoxy resin adhesive. (B) LONG-TERM BEHAVIOR OF STRUCTURES STRENGTHENED BY EPOXY-BONDED STEEL PLATES
Epoxy resins are a new component in the strengthened structural system compared with the original structure made of only concrete and steel. Epoxy resins are polymers whose long-term behavior such as creep and aging are considerably different from those of concrete and steel. The effects of these material properties on the long-term performance of strengthened structures must be considered. The creep of epoxy resins is considerably higher than that of concrete. In
Conventional repair and strengthening
241
thin adhesive layers (normally less than 3 mm), the influence of creep is restricted by the cohesion. For the sake of good long-term performance, the epoxy resin is required to be stiff enough not to creep significantly under sustained load, but flexible enough so that no high stress concentrations can arise (Mays 1985). Aging is a change of properties resulting from long-term mechanical, physical and chemical influences, e.g. relative humidity, radiation, heat, weathering and moisture. For strengthening with steel plate, aging reduces strength so that the long-term strength is only 50 percent of the short-term strength. Epoxy resin adhesives have a certain porosity, which will allow the penetration of water and other solutions. Exposure to water over long periods of time can cause epoxy resin adhesives to lose strength. It is generally required that the epoxy resin for bonding steel-plate has long-term durability for a service life of at least 30 years at service temperatures from −20 °C to +40 °C (Mays 1985). Also, epoxy is generally much more sensitive to fire than concrete so that this strengthening technique is only suitable for concrete structures with low fire risk. In cases of high fire risk, fire protection should be applied to the epoxy as well as steel plate. Some tests show that the fatigue strength of epoxy is about 50 percent of the short-term strength. The fatigue strength, rather than the short-term strength, should be considered when designing the epoxy-bonded steel plates strengthening method for long-term application. 4.4.5 Externally bonded fiber reinforced sheets One of the problems with epoxy-bonded steel plates is that the steel plates are very heavy and sensitive to the moist environment (such as the corrosion damage). There are many types of fiber reinforced materials that have been developed for strengthening applications of reinforced concrete structures, such as glass fiber reinforced polymers (GFRPs) and carbon fiber reinforced polymers (CFRPs) (see ACI 440.2R-02). These fiber reinforced materials are in the form of thin sheets and can be externally bonded onto the existing concrete structure and enhance significantly the load-carrying capacity of the structures. In addition to very high tensile strength of the fibers, GFRPs and CFRPs are much lighter than steel and have very high corrosion resistance. The repair and strengthening applications of GFRPs and CFRPs will be discussed in greater detail in Chapter 5. Similar to CFRPs and GFRPs, steel cord reinforced polymers (SCRP) and steel cord reinforced grout (SCRG) are new materials (Huang et al. 2004) that have found their applications in structural strengthening. SCRP and SCRG are made of thin high-strength steel fibers bundled into cords. The cords are then woven into unidirectional sheets with synthetic textiles. These steel fiber reinforced sheets can be glued onto a concrete structure by either a polymer or a grout similar to the operations for CFRPs or GFRPs. SCRP and SCRG combine the advantages of steel plates and CFRP. They are not as
242
Conventional repair and strengthening
heavy as steel plates and they are covered by polymers or grouts in the application so that the corrosion of the steel is nor longer a problem. Furthermore, the material cost is relatively low compared to that of CFRP; and ductility of the materials is higher than that of carbon and glass fibers such that the strengthened structures have a better performance under excessive dynamic loadings.
5
Glass fiber reinforced plastics components for bridge deck replacement
5.1 Introduction The deterioration of reinforced concrete bridge decks is a major problem in cold regions where salt is commonly employed for de-icing the road surface. The penetration of chloride ions depassivates the steel reinforcement and initiates corrosion. Subsequent water absorption and expansion of the rust lead to cracking and surface spalling of the concrete deck. Moreover, with salt on the road surface, an osmotic pressure will be set up to draw water towards the top of the deck. Once the upper surface reaches a high level of saturation (above 91 percent), the increase in volume associated with freezing water will also lead to the formation of cracks in the concrete. With repeated freezing and thawing, the deck surface can be severely damaged. With the use of conventional repair methods, such as patching and placement of additional steel reinforcements, the functionality and load capacity of the deck can be recovered but the degradation problem is not eliminated. In recent years, a new approach to solving the problem has been developed. When a concrete deck is found to be significantly deteriorated, it is replaced by glass fiber reinforced plastics (GFRP) components, which are prefabricated in the factory and transported to the site for assembly. Since both the glass fiber and the polymeric matrix employed for making GFRP components are non-corroding in the presence of salt, and GFRP is not vulnerable to freezing and thawing, the durability of the GFRP deck is expected to be greatly improved over its concrete counterpart. Moreover, as the strength to weight ratio of GFRP is much higher than that of reinforced concrete, the total weight of the GFRP deck is significantly lower. As a result, the dead load acting on the supporting structure (including the beam girders and columns) is reduced. Such a reduction in dead load is particularly advantageous for composite bridges with steel girders and a concrete deck that have been in service for many years. After years of repeated loading, the loadcarrying capacities of the steel members, as well as that of the connections, are likely to be reduced. By decreasing the total load acting on the supporting members, their serviceable lifetime under further traffic loading can be extended. It should be noted that the relatively low weight of GFRP
244
Glass FRP for bridge deck replacement
members also facilitates the ease of construction. Therefore, while GFRP components are currently used for the replacement of concrete deck in existing bridges, they are also applicable to new bridges under severe environmental conditions. The field application of FRP bridge decks started in 1997. Since then, this technology has been adopted in many projects. An excellent summary of FRP deck applications can be found in a report by Keller (2001).
5.2 Materials Various kinds of glass fibers are available commercially for different performance requirements. Due to cost considerations, E-glass is commonly used to make GFRP components for construction applications. For the polymer, polyester and vinylester are commonly employed. Their advantages include low viscosity at uncured stage and short curing time. Their major disadvantage is the high volumetric shrinkage on curing, which may induce significant residual stresses. Epoxy resin shrinks less during curing, but is not used due to its high cost and long curing time (up to several hours). The general properties of the fiber and matrix materials are summarized in Table 5.1. From Table 5.1, it is clear that the glass fiber has much higher tensile strength and modulus than both polyester and vinylester. Also, in a typical GFRP, the fiber volume fraction ranges from 30 to 60 percent. As a result, the strength and stiffness along the fiber direction are governed by the fibers. The compression strength of glass fiber is not given, due to the fact that single fibers (or fiber bundles) buckle easily under compressive load. When embedded inside a polymer, the composite can carry significant compression and failure occurs through the formation of a shear band by collaborative fiber buckling rather than crushing of the glass fiber. The compressive strength of the glass fiber itself is therefore not a relevant material Table 5.1 Properties of the fiber and matrix materials for the fabrication of GFRP
Fiber diameter (µm) Density (kg/m3) Young’s modulus (GPa) Tensile strength (MPa) Strain at tensile failure (%) Compressive strength Coefficient of expansion (°C) Thermal conductivity (W °C/m) Shrinkage on curing (%) Glass transition temperature (°C) Water absorption 24 hrs at 20 °C (%) Cost (U.S.$/kg)
Glass fiber
Polyester
Vinylester
8–14 2,560 76 1,400–2,500 1.8–3.2 – 4.9 × 10−6 1.04 – – – 2–3
– 1,200–1,500 2–4.5 40–90 2 90–250 1–2 × 10−4 0.2 4–8 70–120 0.1–0.3 2–3
– 1,030–1,170 3.1–3.7 64–89 3–6 116–295 3.2–3.8 × 10−5 0.25 1–8 60–195 0.1–0.3 ~4
Glass FRP for bridge deck replacement
245
parameter to be measured. From Table 5.1, one can also note the large difference in thermal expansion coefficient between the fiber and matrix. However, with the low modulus of polymers relative to glass, the thermally induced stress within the composite is not high. Also, with proper coupling agents applied on the glass surface, good bonding with polymer can be achieved, so debonding will not occur at the end of fibers where high shear stresses are required for strain compatibility to be achieved. In the fabrication of GFRP, both polyester and vinylester are commonly employed. Vinylester is often chosen over polyester for its better chemical and thermal resistance. It also has low viscosity and short curing time but the volumetric shrinkage can be high. One important consideration in the use of GFRP is the reduction of load carrying capacity under sustained loading or cyclic fatigue. For glass fiber alone, the phenomenon of stress rupture (i.e. drop in static strength with the duration of loading) is well known. This effect is attributed to the action of water moisture at the tip of surface cracks. Interaction between water and glass molecules in highly stressed regions results in bond breakage and propagation of cracks. When the crack size increases under sustained loading to reach the critical value, fracture will occur. According to this mechanism, static fatigue should be less of a problem in a dry environment, and this has indeed been experimentally verified. In other words, by limiting the contact of glass with moisture, the static fatigue problem can be alleviated. In GFRP, fibers are embedded inside a polymeric matrix that forms a barrier to water moisture penetration. However, according to many results in the literature, the static fatigue phenomenon is still present in GFRP components. This is an indication that the polymer can slow down the penetration of water but cannot stop it completely. To account for the stress rupture of GFRP members, the design strength is often taken to be much lower (e.g. at about 20 percent) than the value measured in the laboratory. Under cyclic loading conditions, due to the low stiffness of the matrix, the glass fibers carry most of the loading. The fatigue phenomenon observed for single fiber is hence also expected to be found for the GFRP. To design GFRP components under cyclic loading, relevant strength reduction factors should be adopted.
5.3 Fabrication process and example systems Various processes have been developed for the fabrication of GFRP components. In this section, we’ll describe four processes that are particularly suitable for making bridge deck components. 5.3.1 Pultrusion Pultrusion is most suitable for the mass production of components with uniform cross section. Figure 5.1 shows the process with dry fibers as the
246
Glass FRP for bridge deck replacement
Figure 5.1 The pultrusion process for composite fabrication.
starting material. Fibers from spools are first pulled through a resin bath so they are soaked by the polymeric resin to be employed as a matrix. A pair of combs is often located before and after the bath to ensure proper fiber alignment. The wet fibers then pass through a die (which is often heated to facilitate the flow of resin) to form the required shape of the component. The formed component is pulled slowly through an oven for the curing process to be completed. Finally, the fabricated component is cut into specific lengths to be stored and/or transported. Besides continuous fiber strands, weaved fiber cloth or fiber mat (with random chopped fibers embedded inside a resin) can also be pultruded to form composite members. Actually, various combinations of fiber strand, weaved cloth and mat can be employed to produce components with different performance requirements. Various kinds of pultruded GFRP bridge deck components have been commercially produced and applied in the field. Examples are given in Figure 5.2. The Asset system manufactured by Fiberline, Denmark (Figure 5.2(a)) is the result of a research project funded by the European Community. Components are produced in the form of a parallelogram stiffened by a diagonal plate (which divides the parallelogram into two triangles). The individual members are glued together on site to form the bridge deck. To facilitate member connections, extrusions and recesses are provided on the top, bottom and sides. The Superdeck system by Creative Pultrusion, Inc., U.S.A. (Figure 5.2(b)) and the Duraspan system by Martin Marietta and Creative Pultrusion, Inc., U.S.A. (Figure 5.2(c)) are each composed of two matching components. Each kind of component in the Superdeck system is produced by pultrusion alone. For the Duraspan system, the rectangular part (which is divided into two trapezoids by the inclined stiffener) is pultruded, while the upper plate is bonded on afterwards. In the field, the matching components are glued together to form the complete deck. For the above three components, the total depth is 225 mm or below, making them unsuitable for spanning over long distances. In practice, these GFRP members are used to construct a laterally spanning deck system which
Glass FRP for bridge deck replacement
247
Figure 5.2 Examples of pultruded GFRP components: (a) Asset system; (b) Superdeck system; (c) Duraspan system; (d) ACCS system.
has to be supported by longitudinal girders. Field application of these systems as replacement of the concrete bridge deck can be found in the US and Europe. The ACCS (Advanced Composite Construction System) by Maunsell Structural Plastics, UK (Figure 5.2(d)) offers the flexibility to produce box sections of various sizes that can be placed directly between two sets of piers. In other words, both the longitudinal girder and the laterally spanning deck can be replaced by this system. The system consists of three kinds of pultruded elements, including (1) the panel element, formed by multiple squares (each of 75 × 75 mm in size), with special details at the ends for connection; (2) the joint element, which is a single square to be placed at the connection of horizontal and vertical panels; and (3) the toggle element, which is a solid component for connection purposes. With the proper combination of joint elements (which can be either two-way or three-way) and horizontal/vertical panel elements, single or multiple box sections of various sizes can be produced. Using this system, pedestrian bridges and road bridges have been constructed in the UK.
248
Glass FRP for bridge deck replacement
5.3.2 Filament winding Filament winding is a common process for making composite members in tubular forms. One variation of this process is illustrated in Figure 5.3. The key component of the process is a mandrel or internal form, which is fixed to a rotary fixture. The mandrel can be removed from the composite after the fabrication process is complete, or it can stay with the final product. An example of the latter case is a protruded tubular mandrel with fibers along the pultruding direction alone. With fibers wound around its surfaces, the shear and torsional capacity can be significantly improved. As in pultrusion, fibers are first passed through a resin bath and the ends of the wet fibers are fixed on one side of the mandrel. By setting the mandrel in rotation and moving the fibers (together with the whole resin bath) in a direction along the mandrel’s axis of rotation, the fibers are wound onto the mandrel to form a composite member. As a variation of the above process, it is also possible to wind plain fibers onto the mandrel and then add the resin into the fiber assembly through an impregnation process. In either case, after the composite is formed, curing is required for the polymer resin to gain stiffness and strength. 5.3.3 Hand lay-up Hand lay-up is a very versatile process for the fabrication of composites with different fiber arrangements. It relies on the availability of fiber pre-pregs, which are fiber sheets with aligned fibers already wet with uncured or partially cured resins. These pre-pregs are made by passing fiber strands through a resin bath and then wrapping them onto a big roller to form sheets with fiber aligned along the same direction. During the wrapping process, waxed paper is wrapped onto the roller together with the sheet to avoid individual fiber sheets from sticking together. The pre-pregs prepared in this
Figure 5.3 The filament winding process.
Glass FRP for bridge deck replacement
249
manner are then sold as a roll of composite sheet to the user. The typical thickness of glass fiber pre-preg sheet is between 0.25 to 0.50 mm. To make a composite member, the sheet is cut into the required size with the proper fiber orientation. The individual sheets are then stacked on top of one another, until the required thickness is achieved with the specified fiber arrangement. The process is illustrated in Figure 5.4. One can easily see that hand lay-up is suitable for making parts in the form of a plate. It offers excellent design flexibility as the engineer can specify different fiber orientations in different sheets to achieve the required structural performance. In most structural applications, plates formed by hand lay-up are used together with other components to form a highly effective system. For example, in the production of aircraft wings, composite plates are employed as the surface skins separated by metallic stiffeners to take advantage of the high in-plane stiffness and strength of composite materials. For composite bridge decks, the Manitoba system, developed in Canada, adopted a similar approach. The system, which is illustrated in Figure 5.5, makes use of all the composite fabrication techniques discussed so far. Hand-laid plates are placed on the top and bottom, separated by filament wound triangular tubes in the middle. In the wrapping process, to avoid excessively large curvature of the fibers, the corners of the triangular sections are rounded. Pultruded solid rods are added at the corner locations to prevent the formation of voids and
Figure 5.4 The hand lay-up process.
Figure 5.5 The Manitoba deck system.
250
Glass FRP for bridge deck replacement
to further strengthen the member. The Manitoba system is again designed as a laterally spanning deck and needs to be supported by longitudinal girders. 5.3.4 Vacuum-assisted resin transfer molding (VARTM) Vacuum-assisted resin transfer molding (VARTM) is a relatively new composite fabrication method. As illustrated in Figure 5.6, it involves the use of a mold in which plain fibers are placed in a prescribed arrangement. On one side of the mold is an opening connecting to a resin container, while an opening on the other side is connected to a vacuum pump. By drawing a vacuum, resin will flow slowly into the mold and impregnate the fiber. The major advantage of this technique is the shape flexibility, as components of complex shapes can be produced. A modified VARTM method has been used by Hardcore Composites USA to fabricate GFRP components for bridge deck replacement. The deck, as illustrated in Figure 5.7, consists of GFRP plates on the top and bottom, separated by foam material in between. The foam core of the deck is made with hard foam blocks that are wrapped with fiber composites. In the fabrication of the deck, hard foam blocks are placed side by side in between the upper and lower layers of plain glass fibers. After the whole deck is assembled, it is sealed within a bag, and vacuum is drawn to facilitate resin impregnation. In this method, members over 9 m in span have been produced for use as longitudinally spanning deck without the need for additional girders.
Figure 5.6 Illustration of the VARTM.
Figure 5.7 The Hardcore Composite system.
Glass FRP for bridge deck replacement
251
5.3.5 Connections for FRP deck systems In the examples given in Figure 5.2 and Figure 5.5, the connection of adjacent FRP members is illustrated. In practice, the FRP deck system has to be properly secured to the bridge girders or piers. Also, safety rails have to be constructed on the sides of the deck. Typical connection methods and details are provided by the deck manufacturers. However, since typical designs may not cover all site conditions, it is highly advisable for the structural engineer to work together with the manufacturer to come up with the most effective connections that are also easy to implement.
5.4 Analysis of FRP bridge deck members As shown in the last section, most practical FRP bridge deck members are in the form of a multiple tube spanning along one direction or composite plates separated by a foam core. In the former case, as each plate element constituting the tube member has a thickness much smaller than its width, the FRP tube can be treated as a thin-walled structural component, the analysis of which is covered in texts on mechanics of materials. In the latter case, the FRP deck can be analyzed as a sandwich construction with stiff plate elements separated by a soft core. Compared to conventional members made with homogeneous materials, the additional complexity is that each plate element may be made up of many fiber layers with different fiber orientations (Figure 5.8). In other words, the properties of the wall vary over its thickness according to the local fiber arrangement. To analyze the behavior of the FRP member, the equivalent properties of each plate element needs to be obtained first. The theory for finding the equivalent stiffness of a layered composite, as well as the stresses in each layer when loading is applied, is referred to as the classical lamination plate theory. In the following sections, a summary of this theory will be given. Readers interested in further details should refer to the classical text-books by Halpin (1984) and Jones (1998).
Figure 5.8 The composition of an FRP structural member.
252
Glass FRP for bridge deck replacement
Before describing the theory, a number of terminologies and notations need to be introduced. First, a layer of material with fibers aligning along the same direction is called a lamina. A composite made up of a number of bonded lamina is called a laminated composite or simply a laminate. Conventionally, laminated composites are often constructed with pre-pregs using the hand lay-up process. Each layer of pre-preg in the composite (which is technically a lamina) is referred to as a ply. The notation of the composition of a laminated composite follows the fiber orientation in each ply. For example, [0/0/30/−30/90/90/−30/30/0/0] represents a composite built with 10 plies of a particular lamina, with fibers in each layer oriented at the stated angles. As one can see, such a notation can get very lengthy. To simplify the notation, the following is commonly performed: (1) n subsequent plies with the same fiber orientation θ are denoted as θn; (2) two subsequent plies with opposite fiber orientations (+θ and −θ) are grouped together as +θ; (3) symmetrical layer arrangements (which are commonly found in practice for a reason to be explained later) are written as [. . .]s. With these simplifying approaches, the above composite is denoted as [02/+30/90]s. 5.4.1 Elastic behavior of composite lamina The composite lamina comprises aligned fibers in a polymeric matrix. As it is in the form of a thin sheet, we are mainly interested in its in-plane properties. Following the axis notation in Figure 5.9, the in-plane strain and stress components are related by:
⎡1 ⎢ E1 ⎡ ε1 ⎤ ⎢ −ν12 ⎢ ε2 ⎥ = ⎢ E ⎣ γ12⎦ ⎢ 1 ⎢0 ⎣
−ν21 E2 1 E2 0
⎤ ⎥ ⎥ ⎡ σ1 ⎤ 0 ⎥ ⎢ σ2 ⎥ ⎥ ⎣ τ12 ⎦ 1 ⎥ G12 ⎦ 0
(5.1)
where νij = −εj /εi when there is direct stress along the i-direction alone. Also, due to symmetry of the elastic matrix, ν21/E2=ν12 /E1. The elastic constants in Eq (5.1) have similar physical meaning to Young’s modulus and Poisson’s ratio in isotropic materials. Ei and νij can be obtained by applying loading along the i-direction alone, and measuring the strain along both the i and j directions. To find G12, one can either perform the inplane shear test, or the torsion test on a thin cylinder made of the lamina, with the fibers aligning along the axis of the cylinder. In the practical analysis of a component under loading, the strain components often need to be calculated first (see example on p. 259). It is hence necessary to express the stress components in terms of the strain components as:
Glass FRP for bridge deck replacement
253
Figure 5.9 Local (1–2) coordinates for the lamina.
⎡ σ1 ⎤ ⎡ Q11 Q12 0 ⎤ ⎡ ε1 ⎤ ⎡ε1 ⎤ ⎢ σ2 ⎥ = ⎢ Q12 Q22 0 ⎥ ⎢ ε2 ⎥ = [Q] ⎢ε2 ⎥ ⎣ τ12 ⎦ ⎣ 0 0 Q66⎦ ⎣ γ12⎦ ⎣γ12 ⎦
(5.2)
where: Q11 = Q22 =
E1 ; (1 − ν12ν21) E2 (1 − ν12ν21)
;
Q12 =
ν12E2 ν21E1 = ; (1 − ν12ν21) (1 − ν12ν21)
Q66 = G12
It should be noted that the subscript “66” for the shear stiffness arises from the condensation of the 6 × 6 stiffness matrix in three dimensions to the 3 × 3 matrix in two dimensions. We retain this subscript to maintain consistency with other texts on composite laminate theory. 5.4.1.1 Stress–strain relation for a lamina at arbitrary orientation In a composite laminate, the individual layers of lamina are oriented at different directions to achieve the required properties for a particular application. To obtain the “averaged” elastic behavior of the laminate in a given coordinate system, it is necessary to know the transformed properties of each lamina in that system. In Figure 5.10, the 1–2 coordinates correspond to the directions parallel and perpendicular to the fibers, while the x–y axis denotes the “global” coordinates of the composite laminate. For example, for a [0/+30]s laminate, the fibers in the 0-degree lamina is along the x-direction.
254
Glass FRP for bridge deck replacement
Figure 5.10 Definition of local (1–2) and global (x–y) coordinates.
The transformed elastic properties can be derived from stress and strain transformation rules as follows. The stress components in the x–y coordinates are related to those in the 1–2 coordinates by:
⎡ σx ⎤ ⎢ σy ⎥ = ⎣ τxy ⎦
sin2θ − 2 sinθ cosθ ⎤ ⎡cos2θ 2 2 sin θ cos θ 2 sinθ cosθ ⎥ ⎢ sinθ cosθ − sinθ cosθ cos2θ − sin2θ ⎦ ⎣
⎡σ1 ⎤ ⎡ σ1 ⎤ −1 σ = [T] ⎢ 2⎥ ⎢ σ2 ⎥ (5.3a) τ ⎣ 12 ⎦ ⎣ τ12⎦
Similarly,
⎡ εx ⎤ ⎡ ε1 ⎤ ⎢ εy ⎥ = [T]−1 ⎢ ε2 ⎥ ⎣ γxy /2⎦ ⎣ γ12 /2⎦ ⎡ 1 0 0⎤ ⎡ ε1 ⎤ ⎡ ε1 ⎤ ⎢ 0 1 0 ⎥ , we have ⎢ ε2 ⎥ = [R]⎢ ε2 ⎥ ⎣ 0 0 2⎦ ⎣ γ12 ⎦ ⎣ γ12/2⎦ ⎡ εx ⎤ ⎡ εx ⎤ and ⎢ εy ⎥ = [R] ⎢ εy ⎥ . ⎣ γxy ⎦ ⎣ γxy /2⎦
Defining [R] =
(5.3b)
(5.4a)
(5.4b)
From Eq. (5.2) and the transformation of strain components:
⎡ σ1 ⎤ ⎡ ε1 ⎤ ⎡ ε1 ⎤ ⎡ εx ⎤ ⎢ σ2 ⎥ = [Q] ⎢ ε2 ⎥ = [Q][R] ⎢ ε2 ⎥= [Q][R][T] ⎢ εy ⎥ ⎣ τ12 ⎦ ⎣ γ12⎦ ⎣ γ12 /2⎦ ⎣ γxy /2 ⎦ ⎡ εx ⎤ = [Q][R][T][R]−1 ⎢ εx ⎥ ⎣ γxy ⎦
(5.5)
Glass FRP for bridge deck replacement
255
From stress transformation,
⎡ σx ⎤ ⎡ εx ⎤ ⎢ σy ⎥ = [T]−1[Q][R][T][R]−1 ⎢ εy ⎥ = ⎣ τxy ⎦ ⎣ γxy ⎦
–
–
–
⎡Q Q Q ⎤ ⎡ εx ⎤ – 11 – 12 – 16 ⎢Q Q22 Q26 ⎥ ⎢ εy ⎥ 21 – – – ⎣ Q61 Q62 Q66 ⎦ ⎣ γxy⎦
(5.6)
where: – Q11 = Q11 cos4θ + 2(Q12 + 2Q66) sin2θ cos2θ + Q22 sin4θ – – Q12 = Q21 = (Q11 + Q22 − 4Q66) sin2θ cos2θ + Q12(sin4θ + cos4θ) – Q22 = Q11 sin4θ + 2(Q12 + 2Q66) sin2θ cos2θ + Q22 cos4θ (5.7 a–f) – – 3 Q16 = Q61 = (Q11 − Q12 − 2Q66) sinθ cos θ + (Q12 − Q22 + 2Q66) sin3θ cosθ – – Q26 = Q62 = (Q11 − Q12 − 2Q66) sin3θ cosθ + (Q12 − Q22 + 2Q66) sinθ cos3θ – Q66 = (Q11 + Q22 − 2Q12 − 2Q66) sin2θ cos2θ + Q66(sin4θ + cos4θ) – Alternatively, the components of the [Q] matrix can be expressed in terms of the invariants Uis of a given lamina as: – Q11 = U1 + U2 cos 2θ + U3 cos 4θ – – Q12 = Q21 = U4 − U3 cos 4θ – Q22 = U1 − U2 cos 2θ + U3 cos 4θ – – Q16 = Q61 = ½U2 sin 2θ + U3 sin 4θ – – Q26 = Q62 = ½U2 sin 2θ − U3 sin 4θ – Q66 = U5 − U3 cos 4θ
(5.8 a–f)
with U1 = 1/8(3Q11 + 3Q22 + 2Q12 + 4Q66) U2 = ½(Q11 − Q22) U3 = 1/8(Q11 + Q22 − 2Q12 − 4Q66) U4 =
1/8(Q
11
(5.9 a–e)
+ Q22 + 6Q12 − 4Q66)
U5 = 1/8(Q11 + Q22 − 2Q12 + 4Q66) 5.4.2 Classical lamination theory The classical lamination theory covers the analysis of composite laminates under combined effects of bending, torsion and in-plane forces. It is based on Kirchhoff’s assumption of “plane section remaining plane”, and thus is
256
Glass FRP for bridge deck replacement
applicable only to plates with a large ratio of span/thickness. Due to the small thickness of common composite laminates in practice, this condition is normally satisfied. With Kirchhoff’s kinematic assumption, the basic unknowns in the problems are the in-plane displacements (which can be related to the in-plane strains) and the curvatures of the laminate. Let us define uo, vo an wo to be the displacements at the neutral plane of the laminate along the x, y and z directions respectively. From Figure 5.11, the displacement of the laminate along the x-direction (u), at any distance z from the neutral plane, is given by: u = uo − z
∂wo ∂x
(5.10a)
Similarly, the displacement v along the y-direction is: v = vo − z
∂wo ∂y
(5.10b)
The in-plane strain components are then:
⎡ εx ⎤ ⎡ ∂u/∂x ⎤ ⎡ ∂uo /∂x ⎤ ⎡−∂2wo /∂2x ⎤ ⎢ εy ⎥ = ⎢ ∂v/∂y ⎥ = ⎢ ∂vo /∂y ⎥ + z ⎢−∂2wo /∂2x ⎥ ⎣ γxy ⎦ ⎣∂u/∂y + ∂v/∂x ⎦ ⎣∂uo/∂y + ∂vo /∂x⎦ ⎣−∂2wo/∂x∂y ⎦ ⎡ εox ⎤ ⎡κx ⎤ = ⎢ εoy ⎥ + z ⎢κx ⎥ ⎣ γoxy ⎦ ⎣κxy ⎦
(5.11)
where κx, κy and κxy represent the curvatures of the plate. To find the governing equations for the laminate, the notations in Figure 5.12 will be followed. In Figure 5.12, an element of the laminate is shown.
Figure 5.11 Illustration of the plane-section-remains-plane assumption.
Glass FRP for bridge deck replacement
257
Figure 5.12 Definition of forces and bending/torsional moments for the plate.
Nx, Ny and Nxy are the in-plane normal and shear forces per unit length along the sides of the element, Mx, My and Mxy represent the bending and torsional moments per unit length. From equilibrium, the forces and moments can be expressed in terms of stress components by the following integrals over the total thickness (t) of the laminate:
⎡ Nx ⎤ t/2 ⎢ ⎥ ⎢ Ny ⎥ = 冮 ⎢ ⎥ −t/2 N xy ⎣ ⎦
⎡ σx ⎤ ⎢ ⎥ ⎢ σy ⎥ dz ⎢ ⎥ ⎣ τxy ⎦
(5.12a)
⎡ Mx ⎤ t/2 ⎢ ⎥ ⎢ My ⎥ = 冮 ⎢ ⎥ −t/2 ⎣ Mxy ⎦
⎡ σx ⎤ ⎢ ⎥ ⎢ σy ⎥ zdz ⎢ ⎥ ⎣ τxy ⎦
(5.12b)
For a laminate comprising N plies of lamina, each integral in Eq. (5.12) can be written as the summation of N separate integrals, each evaluated over a specific layer. Eq. (5.12a) and (5.12b) can then be combined with Eq. (5.6) and Eq. (5.11) to give:
⎡ Nx ⎤ z ⎢ ⎥ N ⎢ Ny ⎥ = 冱 冮 ⎢ ⎥ k=1 z ⎣ Nxy ⎦ k
k−1
⎡σx ⎤ N ⎢ ⎥ ⎢σy ⎥ dz = 冱 k=1 ⎢ ⎥ τ ⎣ xy ⎦ k
– – – ⎡Q Q12 Q16 ⎤ z 11 ⎢– – – ⎥ ⎢Q21 Q22 Q26 ⎥ 冮 ⎢– – – ⎥z ⎣Q61 Q62 Q66 ⎦ k k
k−1
⎛ ⎡εox ⎜⎢ o ⎜ ⎢εy ⎜⎢ o ⎝ ⎣εy
⎤ ⎡κx ⎤ ⎞ ⎥ ⎢ ⎥ ⎟ ⎥+ ⎢κy ⎥ z ⎟ dz ⎥ ⎢ ⎥ ⎟ ⎦ ⎣κxy ⎦ ⎠ (5.13a)
=
⎡ A11 A12 A16 ⎤ ⎡ ε ⎤ ⎡B11 B12 B16⎤ ⎡κx ⎤ ⎢ A21 A22 A26 ⎥ ⎢ ε ⎥ + ⎢B21 B22 B26⎥ ⎢κy ⎥ ⎣ A61 A62 A66 ⎦ ⎣ γ ⎦ ⎣B61 B62 B66⎦ ⎣κxy ⎦ o x o y o xy
Glass FRP for bridge deck replacement
258
⎡ Mx ⎤ z ⎢ ⎥ N ⎢ My ⎥ = 冱 冮 ⎢ ⎥ k=1 z ⎣ Mxy⎦ k
k−1
– ⎡ σx ⎤ ⎡Q 11 N ⎢ ⎥ ⎢– ⎢ σy ⎥ zdz = 冱 ⎢ Q21 k=1 ⎢ ⎥ ⎢– ⎣ τxy⎦ k ⎣ Q61
– – Q12 Q16 ⎤ – – ⎥ Q22 Q26 ⎥ – Q62
– ⎥z Q66 ⎦ k
zk
冮 k−1
⎛ ⎡εox ⎤ ⎡κx ⎤ ⎞ ⎜ ⎢ o ⎥ ⎢ ⎥ 2⎟ ⎜ ⎢εy ⎥ z + ⎢κy ⎥ z ⎟dz ⎜⎢ o⎥ ⎢ ⎥ ⎟ ⎝ ⎣γxy⎦ ⎣κxy ⎦ ⎠
⎡ B11 B12 B16 ⎤ ⎡ εox ⎤ ⎡D11 D12 D16⎤ ⎡ κx ⎤ = ⎢ B21 B22 B26 ⎥ ⎢ εoy ⎥ + ⎢D21 D22 D26⎥ ⎢ κy ⎥ ⎣ B61 B62 B66 ⎦ ⎣ γoxy ⎦ ⎣D61 D62 D66⎦ ⎣ κxy⎦
(5.13b)
where: N
Aij =
冱 (Q– )
ij k
(zk − zk − 1);
k=1
N
1
Bij = 2
冱 (Q– )
ij k
(z2k − z2k − 1);
k=1 N
Dij = 31
冱 (Q– )
ij k
(z3k − z3k − 1)
k=1
Eq. (5.13a) and Eq. (5.13b) can be combined together to form a 6 × 6 matrix:
冤M˜冥 = 冤B D冥 冤 κ˜ 冥 Ñ
A
B
ε˜o
(5.14)
with: Ñ being a 3 × 1 vector representing the forces/unit length; ˜ being a 3 × 1 vector representing the moments/unit length; M ε˜o being a 3 × 1 vector representing the in-plane strain components along the mid-plane of the laminate; κ˜ being a 3 × 1 vector representing the curvatures of the laminate. Also, [A], [B] and [D] are the 3 × 3 matrices in Eqs (5.13a) and (5.13b). Eq. (5.14) is the governing equation to find the strains and curvatures associated with a given set of applied forces and moments. Due to the presence of the sub-matrix [B] in the equation, there is plausible coupling between the in-plane deformation and the bending/torsional moments as well as the curvatures and the in-plane forces. In other words, if pure
Glass FRP for bridge deck replacement
259
bending is applied to the laminate, there may be elongation, shortening or shearing of the mid-plane. Similarly, under the presence of in-plane forces alone, the laminate may undergo bending or twisting. From the design engineer’s point of view, this kind of coupling is not desirable, as any restraint to the induced deformation will induce secondary stresses into the member. It is therefore common practice to arrange the lamina layers in such a way as to make all the components of the [B] matrix zero. A simple approach is to use a symmetric laminate. For each layer at a given orientation above the mid-plane, there is a corresponding layer of the same orientation at the same distance below the mid-plane. The contribution of these two layers to any component of the [B] matrix will then sum to zero (the proof is left to the reader as an exercise). With zero contribution from each pair of lamina above and below the mid-plane, the [B] matrix becomes the null matrix and the coupling effects described above disappear. In the remaining part of the present chapter, we will only consider symmetric laminates which are commonly employed in practice. For symmetric laminates, there is still the possibility of coupling between the direct stress and shear strain through the A16 and A26 components. In other words, when direct tension (or compression) is applied, shear deformation is also resulted. To eliminate this kind of coupling, the laminate should be designed with each angled lamina accompanied by a lamina at opposite angle (i.e. for any number of +θ lamina, there should be an equal number of −θ lamina). Also, through the D16 and D26 components, a bending moment will result in the twisting of the member. Since it is impossible to design a laminate with D16, D26, and all the components of the [B] matrix to be zero at the same time, the bending/twisting coupling is allowed to remain in the practical design of composite laminates. With Eq. (5.14), the strain components ε˜o at the middle of the composite laminate as well as the curvatures κ˜ can be calculated from the applied force and moments/torsion. Knowing ε˜o and κ˜, the in-plane strain components in each lamina layer (which vary linearly over the thickness of the lamina) can – be obtained. Then, using the corresponding [Q] matrix for each lamina, the stress components are calculated. To illustrate the calculation procedures, an example is given below. Example 5.1 A [02/±30]s laminate is constructed with GFRP lamina with the following properties: E1 = 53.8 GPa, E2 = 17.9 GPa, ν12 = 0.25, G12 = 8.6 GPa, Ply thickness = 0.25 mm Find the in-plane stress components in each ply under:
260 1 2
Glass FRP for bridge deck replacement in-plane forces Nx =20,000 MN/m; Moment Mx = 200 Nm/m.
The first step in solving this problem is to derive the [A] and [D] matrices. From the given elastic properties, ν21 =
ν12E2 = 0.08318 E1
Q11 =
E1 (1 − v12v21)
= 54.94 GPa
Q12 =
v21E1 v12E2 = = 4.57 GPa (1 − v12v21) (1 − v12v21)
Q22 =
E2 = 18.28 GPa (1 − v12v21)
Q66 = G12 = 8.6 GPa The invariants are given by: U1 = 81 (3Q11 + 3Q22 + 2Q12 + 4Q66) = 32.90 GPa U2 = 12 (Q11 − Q22) = 18.33 GPa U3 = 18 (Q11 + Q22 − 2Q12 − 4Q66) = 3.71 GPa U4 = 81 (Q11 + Q22 + 6Q12 − 4Q66) = 8.28 GPa U5 = 18 (Q11 + Q22 − 2Q12 + 4Q66) = 12.31 GPa Each component of the [A] matrix can be calculated from: N
Aij =
冱 (Q– ) (z ij k
k
− zk − 1)
k=1
with (zk − zk − 1) being the thickness of each ply which is 0.25 mm. According to Eq. (5.8), the above summation involves the summation of cos 2θ, cos 4θ, cos 2θ and cos 2θ over all the plies. This can be performed systematically in a table form (Table 5.2). 8
A11 = tply
冱 [U
1
+ U2 cos 2θ + U3 cos 4θ]k
k=1
8
8
= tply [8U1 +
冢 冱 cos 2θ冣U + 冢 冱 cos 4θ冣U ] 3
2
k=1
k=1
= tply [8U1 + 6U2 + 2U3] = 95.15 MN/m
Glass FRP for bridge deck replacement
261
Table 5.2 Summation over all the plies Ply
Angle
cos 2θ
cos 4θ
sin 2θ
sin 4θ
1 2 3 4 5 6 7 8 Sum
0 0 +30 −30 −30 +30 0 0
1 1 0.5 0.5 0.5 0.5 1 1 6
1 1 −0.5 −0.5 −0.5 −0.5 1 1 2
0 0 0.866 −0.866 −0.866 0.866 0 0 0
0 0 0.866 −0.866 −0.866 0.866 0 0 0
With similar calculations, the other components of the [A] matrix are found to be: A12 = 14.71 MN/m A22 = 40.16 MN/m A66 = 22.77 MN/m 8
8
k=1
k=1
冢 冱 sin 2θ = 0 and 冱 sin 4θ = 0冣
A16 = A26 = 0 ⬗
– To find the [D] matrix, it is necessary to determine the Qij component for each ply first. • •
– For the 0° plies: The components of the [Q] matrix is the same as that of the [Q] matrix. For the +30° ply: – Q [+30] = U1 + U2 cos(2 × 30°) + U3 cos(4 × 30°) = 40.21 GPa 11 – [+30] Q 12 = U4 − U3 cos(4 × 30°) = 10.14 GPa – Q [+30] = U1 − U2 cos(2 × 30°) + U3 cos(4 × 30°) = 21.88 GPa 22 – [+30] 1 Q 16 = 2U2 sin(2 × 30°) + U3 sin(4 × 30°) = 11.15 GPa – Q [+30] = 12U2 sin(2 × 30°) − U3 sin(4 × 30°) = 4.72 GPa 26 – Q [+30] = U5 − U3 cos(4 × 30°) = 14.17 GPa 66
•
For the −30° ply: Since cos(−α) = cos(α) and sin(−α) = −sin(α), – – Q [−30] = Q[+30] = 40.21 GPa 11 11 – [−30] – [+30] Q 12 = Q12 = 10.14 GPa
262
Glass FRP for bridge deck replacement – – Q [−30] = Q[+30] = 21.88 GPa 22 22 – [−30] – [+30] Q 66 = Q66 = 14.17 GPa – – Q [−30] = −Q[+30] = −11.15 GPa 16 16 – – Q [−30] = −Q[+30] = −4.72 GPa 26 26
The components of the [D] matrix are then given by: 8
D11 =
1 3
冱 (Q–
) (z3k − z3k − 1)
11 k
k=1
= 2 × 31 [(64 − 27)(54.94) + (27 − 8)(54.94) + (8 − 1)(40.21) + (1 − 0)(40.21)] t 3ply = 35.40 Nm D12 = 2 × 31 [(64 − 8)(4.57) + (8 − 1)(10.14) + (1 − 0)(10.14)]t3ply = 3.51 Nm D22 = 2 × 31 [(64 − 8)(18.28) + (8 − 1)(21.88) + (1 − 0)(21.88)]t3ply = 12.49 Nm D16 = 2 × 31 [(64 − 8)(0) + (8 − 1)(11.15) + (1 − 0)(−11.15)]t3ply = 0.697 Nm D26 = 2 × 13 [(64 − 8)(0) + (8 − 1)(4.72) + (1 − 0)(−4.72)]t3ply = 0.295 Nm D66 = 2 × 31 [(64 − 8)(8.6) + (8 − 1)(14.17) + (1 − 0)(14.17)]t3ply = 6.20 Nm (i) When Nx = 20,000 N/m is applied, the in-plane strains at the mid-plane are given by:
⎡ 95.15 14.71 0 ⎤ ⎡εox ⎤ ⎡20,000 ⎤ ⎢ 14.71 40.16 0 ⎥ ⎢εoy ⎥ = ⎢0 ⎥ × 10−6 o 0 22.77 ⎦ ⎣γxy ⎦ ⎣0 ⎣0 ⎦ Note that the factor 10−6 appears in the above matrix as the Aij components are in MN/m while the applied force is in N/m. On solving,
⎡ εox ⎤ ⎡222.80 ⎤ ⎢ εoy ⎥ = ⎢−81.61 ⎥ × 10−6 ⎣ γoxy ⎦ ⎣0 ⎦ The stress components in each ply can then be calculated.
Glass FRP for bridge deck replacement •
263
For the 0° plies:
⎡ σx ⎤ ⎡54.94 4.57 0 ⎤ ⎡22.80 ⎤ ⎢ σy ⎥ = ⎢4.57 18.28 0 ⎥ (× 103MPa) ⎢−81.61 ⎥ × 10−6 ⎣ τxy ⎦ ⎣0 0 8.6 ⎦ ⎣0 ⎦ ⎡ 11.87 ⎤ = ⎢ −0.474⎥ (MPa) ⎣0 ⎦ •
For the +30° plies:
⎡ σx ⎤ ⎡ 40.21 10.14 11.15 ⎤ ⎡222.80⎤ ⎡8.13 ⎤ ⎢ σy ⎥ = ⎢ 10.14 21.88 4.72 ⎥ × 103 ⎢−81.61⎥ × 10−6 = ⎢−0.474⎥ (MPa) ⎣ τxy ⎦ ⎣ 11.15 4.72 14.17 ⎦ ⎣0 ⎦ ⎣2.10 ⎦ •
For the −30° plies:
⎡ σx ⎤ ⎡ 40.21 10.14 −11.15⎤ ⎡ 222.80⎤ ⎢ σy ⎥ = ⎢ 10.14 21.88 −4.72 ⎥ × 103 ⎢ −81.61⎥ × 10−6 ⎣ τxy ⎦ ⎣ −11.15 −4.72 14.17⎦ ⎣0 ⎦ ⎡8.13 ⎤ = ⎢0.474 ⎥ (MPa) ⎣−2.10 ⎦ The stress distribution along the upper 4 plies of the laminate is shown in Figure 5.13(a). Due to symmetry, the stresses in the lower 4 plies are mirror images of those in the upper plies. One can take a closer look at the calculated results to see if they are consistent with physical intuition. First of all, the stresses must satisfy global equilibrium. For example, N
冱t
σ = 0.25 × 10−3[(11.87 + 11.87 + 8.13 + 8.13) × 106] × 2
ply x
k=1
= 20,000 N/m Along the x-direction, the 0° ply is stiffer than the 30° ply. Under the same value of εx, the stress is higher for the 0° ply. Also, as both the +30° and −30° plies exhibit the same stiffness, they carry the same σx. With fibers at 30° to the loading direction, rotation of fibers under the applied loading leads to increased lateral deformation. The Poisson’s effect is hence more significant than the 0° ply. To achieve compatibility, lateral compressive stresses are induced in the 0° ply while the 30° ply is under lateral tension. This explains the presence of σy in all the plies, as well as the opposite sign of the stresses in the 0° and 30° plies.
264
Glass FRP for bridge deck replacement
The application of Nx on the laminate does not produce any shear strain. For the 0° ply, the shear stress is also zero. For the +30° and −30° plies, shear stresses of the same magnitude but opposite signs are induced. (ii) When Mx = 200 Nm/m is applied on the laminate, the curvatures are given by:
⎡ 35.40 3.51 0.697 ⎤ ⎡ κx ⎤ ⎡200⎤ ⎢ 3.51 12.49 0.295 ⎥ ⎢ κy ⎥ = ⎢0 ⎥ ⎣ 0.697 0.295 6.20 ⎦ ⎣ κxy⎦ ⎣0 ⎦ ⎡ κx ⎤ ⎡ 5.82 ⎤ On solving, ⎢ κy ⎥ = ⎢−1.62 ⎥ /m ⎣ κxy⎦ ⎣−0.58 ⎦ According to Kirchhoff’s assumption for a plate under bending and/or torsion, the strain is linearly distributed along the depth of the plate. The stresses in each ply also vary linearly over its thickness, with a magnitude directly proportional to distance from the mid-plane. In the following, the stresses at the top of the laminate as well as the boundaries between plies of different properties are derived.
Figure 5.13 Stress distribution in the upper half of the [02/±30]s laminate under (a) Nx = 20,000 MN/m, and (b) Mx = 200 Nm/m.
Glass FRP for bridge deck replacement At the top of the laminate
⎡ σx ⎤ ⎡ 54.94 4.57 0 ⎤ ⎡ 5.82 ⎤ ⎢ σy ⎥ = ⎢ 4.57 18.28 0 ⎥ × 103 ⎢−1.62 ⎥ × 4 × 0.25 × 10−3 0 8.6⎦ ⎣ τxy ⎦ ⎣ 0 ⎣−0.58 ⎦ ⎡312.44⎤ = ⎢−3.05 ⎥ (MPa) ⎣−4.96 ⎦ At the boundary between the 0° and +30° plies Right above the boundary, within the 0° ply,
⎡ σx ⎤ ⎡ 312.35⎤ ⎡ 156.22 ⎤ ⎢ σy ⎥ = ⎢ −3.02 ⎥ × 42 = ⎢ −1.53 ⎥ (MPa) ⎣ τxy ⎦ ⎣ 4.558 ⎦ ⎣ −2.48 ⎦ Right below the boundary, within the +30° ply,
⎡ σx ⎤ ⎡ 40.21 10.14 11.15⎤ ⎡ 5.82 ⎤ ⎢ σy ⎥ = ⎢ 10.14 21.88 4.72 ⎥ × 103 ⎢−1.62 ⎥ × 2 × 0.25 × 10−3 ⎣ τxy ⎦ ⎣ 11.15 4.72 14.17⎦ ⎣−0.58 ⎦ ⎡105.61⎤ = ⎢10.41 ⎥ (MPa) ⎣24.54 ⎦ At the boundary between the +30° and −30° plies Right above the boundary, within the +30° ply,
⎡ σx ⎤ ⎡ 105.84⎤ ⎡52.81⎤ ⎢ σy ⎥ = ⎢ 10.53 ⎥ × 12 = ⎢ 5.20⎥ (MPa) ⎣ τxy ⎦ ⎣ 24.87 ⎦ ⎣12.27⎦ Right below the boundary, within the −30° ply,
⎡ σx ⎤ ⎢ σy ⎥ = ⎣ τxy ⎦ =
⎡ 40.21 10.14 −11.15 ⎤ ⎢ 10.14 21.88 −4.72 ⎥ × 103 ⎣ −11.15 −4.72 14.17 ⎦ ⎡ 56.02 ⎤ ⎢ 6.56⎥ (MPa) ⎣−16.36⎦
⎡ 5.82 ⎤ ⎢−1.62 ⎥ × 1 × 0.25 × 10−3 ⎣−0.58 ⎦
265
266
Glass FRP for bridge deck replacement
The distribution of stresses in the upper 4 plies is shown in Figure 5.13(b). As expected, σx is higher in the outer layers because the strains are higher and the 0° plies possess higher stiffness along the x-direction than the 30° plies. The discontinuity in stress between the 0° and +30° plies is mainly due to the stiffness difference. Interestingly, the stress is also discontinuous at the interface between the +30° and −30° plies. This is due to the presence of shear strains arising from the torsional curvature κxy, which have different effects on plies of opposite angles. As an exercise, the reader can show that the sum of moments from stresses in each ply is in equilibrium with the applied moment. For σy and τxy, one can observe sign changes in the stress values. Also, the stresses tend to be smaller for the plies away from the mid-plane. These observations are consistent with physical intuition because the total moments produced by these stresses must sum to zero over all the plies. In this example, the effect of axial force and bending are considered separately. In practice, both may be present simultaneously, together with shear force and torsion. However, as long as the laminate is symmetric (which is the case in most practical situations), so the matrix [B] is zero, the effects of in-plane forces and bending/torsional moments can be considered separately to determine ε˜o and κ˜. The total strain (or stress) in each ply is then obtained by superposition. 5.4.3 Equivalent forces and moments arising from thermal and hygroscopic effects In the above, we have focused on the effect of externally applied forces and moments on the laminate. However, for a composite laminate with different fiber orientations in different plies, the thermal expansion coefficient along any direction will also be different for each ply. As a result, when temperature changes occur, stresses will be induced in the individual plies to maintain compatibility of the laminated composite. This is of particular relevance for laminates made with resins that require a high curing temperature. After the material hardens at the curing temperature, cooling to room temperature will introduce high residual stresses in the plies. Besides temperature changes, variation in moisture content of the composite (due to water absorption by the polymer) also results in dimensional changes. This is known as the hygroscopic effect. The dimensional change is less significant along the fiber direction than along the direction perpendicular to the fiber. Again, with fibers at different orientations in different plies, the hygroscopic effect will result in stresses in each ply. To quantify the thermal and hygroscopic effects (which are often combined together and called the hygrothermal effects), the superposition principle can be employed. One can assume that external forces and moments have been applied to the laminate to resist the deformation so the laminate stays un-deformed. Since these external effects are fictitious and
Glass FRP for bridge deck replacement
267
not present in reality, opposite forces and moments need to be applied to cancel them out. The strains and curvatures arising from the latter set of forces and moments represent the effect of temperature and moisture change. Therefore, once the equivalent forces and moments corresponding to the hygroscopic effects are obtained, the same approach as in the last section can be employed to find the mid-plane strains and curvatures. To find the equivalent forces and moments, it is noted that the strain in each ply induced by temperature and moisture changes can be written as:
⎡ εHx ⎤ ⎢ εHy ⎥ = ⎢ H⎥ ⎣ γxy ⎦
⎡αxΔT + βxΔm ⎤ ⎢αyΔT + βyΔm ⎥ ⎢ ⎥ ⎣αxyΔT + βxyΔm ⎦
(5.15)
where ΔT and Δm are the change in temperature and moisture content respectively; αx, αy and αxy are the transformed thermal expansion coefficients in the x–y coordinate system; βx, βy and βxy are the transformed hygroscopic expansion coefficients. Also, by defining α1 and α2 as the thermal expansion coefficients of the lamina along and perperdicular to the fiber direction, and β1 and β2 as the corresponding hygrosopic expansion coefficients, it can be shown by simple strain transformations that: αx = α1 cos2θ + α2 sin2θ; βx = β1 cos2θ + β2 sin2θ αy = α1 sin2θ + α2 cos2θ; βy = β1 sin2θ + β2 cos2θ αxy = (α1 − α2) sinθ cosθ; βxy = (β1 − β2) sinθ cosθ To prevent the laminate from deformation, fictitious forces and moments are required to produce strains that are equal in magnitude but opposite in sign to the thermal and hygroscopic strains give in Eq. (5.15). In each ply, the required stress is given by:
⎡ σHx ⎤ ⎢ σHy ⎥ = ⎢ H⎥ ⎣ τxy ⎦
– – – ⎡Q ⎡ βx ⎤ ⎞ Q12 Q16 ⎤ ⎛ ⎡ αx ⎤ 11 – – – ⎢ Q21 Q22 Q26 ⎥ ⎜− ⎢ αy ⎥ ΔT − ⎢ βy ⎥ Δm ⎟ ⎢– ⎥ ⎢ ⎥ ⎟ – – ⎥ ⎜ ⎢ ⎣ Q61 Q62 Q66 ⎦ k ⎝ ⎣ αxy ⎦k ⎣ βxy ⎦ k ⎠
(5.16)
268
Glass FRP for bridge deck replacement – – – ⎡ N Hx ⎤ ⎡σHx ⎤ ⎡Q Q12 Q16⎤ 11 z N N ⎢ H⎥ ⎢ H⎥ ⎢– – – ⎥ ⎢N y ⎥ = ⎢σ y ⎥ dz = ⎢ Q21 Q22 Q26⎥ k
⎢ H⎥ ⎣ N xy ⎦
冱冮
k=1 z k−1
zk
冮 zk − 1
⎛ ⎜ ⎜− ⎜ ⎝
⎡ αx
k
k−1
冮 zk − 1
⎛ ⎜ ⎜− ⎜ ⎝
冱
k=1
⎢– – – ⎥ ⎣ Q61 Q62 Q66⎦ k
(5.17a)
⎤ ⎤ ⎞ ⎢ ⎥ ⎢ ⎥ ⎟ ⎢ αy ⎥ ΔT − ⎢βy ⎥ Δm⎟ dz ⎢ ⎥ ⎢ ⎥ ⎟ ⎣ αxy ⎦k ⎣βxy ⎦ k ⎠
⎡ MHx ⎤ z ⎢ H⎥ N ⎢M y ⎥ = 冱 冮 ⎢ H⎥ k = 1 z ⎣ Mxy⎦ zk
⎢ H⎥ ⎣τxy ⎦ k
⎡βx
⎡σHx ⎤ N ⎢ H⎥ σ zdz = ⎢ y⎥ 冱 k=1 ⎢ H⎥ ⎣τxy ⎦ k
– – – ⎡Q Q12 Q16 ⎤ 11 ⎢– – – ⎥ ⎢ Q21 Q22 Q26 ⎥ ⎢– – – ⎥ ⎣ Q61 Q62 Q66 ⎦ k
(5.17b)
⎡ αx ⎤ ⎡βx ⎤ ⎞ ⎢ ⎥ ⎢ ⎥ ⎟ ⎢ αy ⎥ ΔT − ⎢βy ⎥ Δm⎟ zdz ⎢ ⎥ ⎢ ⎥ ⎟ ⎣ αxy ⎦k ⎣βxy ⎦ k ⎠
The equivalent forces and moments are opposite to the fictitious actions applied to maintain zero deformation, and can be written as:
⎡ N eqx ⎤ ⎡ N Hx ⎤ N ⎢ eq ⎥ ⎢ H⎥ ⎢N y ⎥ = − ⎢ N y ⎥ = 冱 ⎢ eq ⎥ ⎢ H⎥ k = 1 N xy ⎣ ⎦ ⎣ N xy⎦ zk
冮 zk−1
– – – ⎡Q Q12 Q16 ⎤ ⎛ ⎡ αx ⎤ 11 ⎢– – – ⎥ ⎜⎢ ⎥ ⎢ Q21 Q22 Q26 ⎥ ⎜ ⎢ αy ⎥ ⎢– – – ⎥ ⎜⎢ ⎥ ⎣ Q61 Q62 Q66 ⎦ k ⎝ ⎣ αxy ⎦ k
⎡ βx ⎤ ⎢ ⎥ ΔTdz + ⎢ βy ⎥ ⎢ ⎥ ⎣ βxy ⎦ k
zk
冮 zk−1
⎞ ⎟ Δmdz ⎟ ⎟ ⎠
(5.18a)
Glass FRP for bridge deck replacement
269
– – – ⎡ M eqx⎤ ⎡ M Hx ⎤ N ⎡ Q Q12 Q16 ⎤ ⎛ ⎡ αx ⎤ 11 ⎢ eq⎥ ⎢ H⎥ ⎢– – – ⎥ ⎜⎢ ⎥ ⎢ M y ⎥ = − ⎢ M y ⎥ = 冱 ⎢ Q21 Q22 Q26 ⎥ ⎜ ⎢ αy ⎥ ⎢ eq⎥ ⎢ H⎥ k = 1 ⎢ – – – ⎥ ⎜⎢ ⎥ M xy ⎣ ⎦ ⎣ M xy⎦ ⎣ Q61 Q62 Q66 ⎦ k ⎝ ⎣ αxy ⎦ k zk
冮 zk−1
⎡ βx ⎤ ⎢ ⎥ ΔTzdz + ⎢ βy ⎥ ⎢ ⎥ ⎣ βxy ⎦ k
zk
冮 zk−1
⎞ ⎟ Δmzdz ⎟ ⎟ ⎠
(5.18b)
Generally speaking, a thermal analysis or moisture diffusion analysis needs to be performed first to obtain the temperature or moisture variation over the thickness of the laminate. Knowing the variation of ΔT and Δm with z (the position along the thickness direction), the above equations can be employed to calculate the equivalent forces and moments. These can then be superposed to the externally applied forces/moments to calculate ε˜o and κ˜ for the laminate. In the presence of hygrothermal effects, an important point should be noted about the calculation of stresses. For each ply, the stress is given by: – – – ⎡ σx ⎤ ⎡Q ⎡αx ⎤ ⎡ βx ⎤ ⎞ Q12 Q16⎤ ⎛ ⎡εox ⎤ ⎡κx ⎤ 11 – – – ⎥ ⎜⎢ o ⎥ ⎢ ⎥ ⎢ σ ⎥ = ⎢Q ⎢ ⎥ ⎢ ⎥ Q Q ε + κ z− α ΔT − βy Δm ⎟ ⎢ y ⎥ ⎢ – 21 – 22 – 26⎥ ⎜ ⎢ yo ⎥ ⎢ y ⎥ ⎢ y ⎥ ⎢ ⎥ ⎟ ⎣ τxy ⎦ k ⎣Q61 Q62 Q66⎦ k ⎝ ⎣γxy ⎦ ⎣κxy ⎦ ⎣αxy ⎦ k ⎣ βxy ⎦ k ⎠ (5.19) The physical meaning of the equation is that the stresses are arising from the difference in final strain and the relative dimensional changes caused by hygrothermal effects alone. 5.4.4 Failure of composite laminates In the above, a framework is presented for the derivation of strains and stresses in a composite laminate under both external loading and hygrothermal effects. After knowing the stresses in the individual plies, one would like to know if failure would occur in any of the layers. To do so, a failure criterion for the lamina is required. In the literature, various failure criteria have been proposed, and several major ones are described below. 5.4.4.1 Maximum stress theory According to this theory, the composite lamina will NOT fail as long as:
270
Glass FRP for bridge deck replacement XC < σ 1 < XT YC < σ2 < YT
(5.20)
⎢τ12 ⎢ < S where: σ1, σ2 and τ12 are the in-plane stress components with respect to the principal directions of the lamina (i.e. the directions parallel and perpendicular to the fibers); XC and XT are the compressive and tensile strength along the fiber direction respectively. (Note: compressive stress is taken to be negative here.) YC and YT are the compressive and tensile strength perpendicular to the fiber direction; S is the shear strength of the laminate. In the analysis of composite laminates, the strain and stress components are first obtained with respect to the global coordinates. For layers with fibers not aligning with one of the global coordinate axes, it is necessary to transform the stresses back to the principal axes of the lamina first, using the transformation matrix [T] in Eq. (5.3). This failure criterion is very simple and does not consider any interaction between direct and shear stresses, or that between σ1 and σ2. The use of this criterion may therefore over-estimate the failure load of the lamina. 5.4.4.2 Maximum strain theory According to this theory, the composite lamina will NOT fail as long as: εC1 < ε1 = εC2 < ε1 =
σ1 − ν12σ2 E1
< εT1
σ2 − ν21σ1 T < ε2 E2
(5.21)
τ ⎢γ12 ⎢ = ⎢⎢ 12 ⎢⎢ < Γ12 G12
where: ε1, ε2 and γ12 are the in-plane strain components with respect to the principal directions of the lamina; ε1C and ε1T are the compressive and tensile failure strain along the fiber direction respectively. (Note: compressive strain is taken to be negative here.)
Glass FRP for bridge deck replacement
271
ε2C and ε2T are the compressive and tensile failure strain perpendicular to the fiber direction; Γ12 is the shear failure strain of the laminate. By considering strain components in the failure criterion, the interaction between stresses along and perpendicular to the fibers is somewhat accounted for. However, the interaction between the direct and shear stresses is still not considered. 5.4.4.3 The Tsai-Hill theory According to this theory, failure of the lamina occurs when: σ21 X
2
−
2 σ1σ2 σ22 τ12 + 2+ 2 =1 2 X Y S
(5.22)
σ1, σ2, τ12 and S are defined as above. X is taken to be XT if σ1 is in tension, and XC if σ1 is in compression. Similarly Y is taken to be YT or YC depending on the sign of σ2. In this failure criterion, the interaction of the various stress components is considered. Experimental results on the failure of lamina with loading applied at various degrees to the fiber direction show that the Tsai-Hill criterion can predict the failure load better than either the maximum stress or maximum strain criterion. One disadvantage of the Tsai-Hill criterion is that the user has to determine which strength to use (e.g. XT or XC) according to the sign of the stresses, which will result in a certain degree of inconvenience in practice. Moreover, if the criterion is plotted in stress space, the surface is not smooth at locations where σ1 or σ2 changes in sign. The presence of discontinuities in the gradient can impose difficulties in numerical analysis. Such a disadvantage is alleviated by the next criterion to be discussed. 5.4.4.4 The tensor polynomial failure criterion According to this criterion, failure occurs when: 2 1 1 1 σ21 σ22 τ12 1 + σ1 + + σ2 − − + 2 =1 XT XC YT YC XTXC YTYC S
冢
冣
冢
冣
(5.23)
Under many practical situations, predictions from this criterion and the Tsai-Hill criterion are very similar. As this criterion gives a smooth surface in the stress space, it is commonly employed when numerical analysis is to be performed.
272
Glass FRP for bridge deck replacement
5.4.4.5 The first-ply failure and post-failure analysis After knowing the stress components in each ply of a composite laminate, application of a lamina failure criterion allows us to tell if failure has occurred in any of the plies. The simplest approach for failure prediction of laminates is to assume ultimate failure to occur at the instant of first-ply failure. In other words, if any of the plies in a laminate fails, the composite is considered to have reached its full load capacity. It is quite obvious that such an approach can be very conservative in many cases. For example, one can consider the situation with tension applied to a laminate consisting of ten 0°-plies and two 90°-plies, along the 0° direction. As the strength of the lamina is much lower at a direction perpendicular to the fibers, the 90°-plies will fail at a rather low load level. However, after the 90°-plies fail, the 0°-plies can continue to support a much higher loading. The first-ply failure approach will then significantly underestimate the load capacity of the composite. To avoid over-conservativeness in some cases, behavior after first-ply failure can be analyzed with updated [A] and [D] matrices calculated with reduced stiffness of the failed ply. If failure in a ply is due to fiber rupture (i.e. dominated by σ1), all the stiffness components of the ply are taken to be zero. In other words, that particular ply is completely removed from the analysis. If ply failure occurs in the matrix instead (i.e. failure dominated by σ2 or τ12), only the stiffness along the fiber is taken to remain. That is, only Q11 is non-zero for that particular ply. Analysis of the composite can be continued with updated matrices after the failure of each ply until the load can no longer be increased. Ultimate failure is then reached. 5.4.4.6 Interlaminar failure In practice, the failure of laminates may not initiate within a particular ply. Instead, adjacent plies in the laminate may separate from one another due to the presence of shear and normal tensile stresses between the plies. A typical interlaminar failure is illustrated in Figure 5.13, which shows two plies with opposite angles delaminating from one another, inducing final failure by shearing along the fibers in each ply. The presence of interlaminar stresses can be explained with a simple example, where direct tension is applied along the x-direction of a laminate (Figure 5.14). Due to different fiber
Figure 5.14 Interlaminar failure between opposite angle plies.
Glass FRP for bridge deck replacement
273
Figure 5.15 Physical explanation for the presence of interlaminar shear stresses.
orientations in each ply, the Poisson’s ratio of a particular ply (along the loading direction) can be very different to that of an adjacent ply. As shown in Example 1, such a mismatch in Poisson’s ratio can induce in-plane stresses in the y-direction, which are required for displacement compatibility to be maintained. However, as shown in Figure 5.15, σy must be zero along the lateral surface of the laminate. In order for non-zero σy to appear in a particular ply, force equilibrium requires the presence of shear stresses along one or both interfaces between the ply and its adjacent plies. In practice, σy builds up very quickly and reaches the required value in a certain ply at a small distance from the lateral side. The shear stresses are hence very high near the side, but decreases rapidly to zero. The presence of normal interlaminar stresses is related to the shear stresses. As one can imagine, the interlaminar shear stresses on the top and bottom of a certain ply may not produce the same resultant force. As a result, the moment produced by the corresponding shear forces are not balanced, and will cause the ply to bend (in the y–z plane). In general, different plies in the laminate may bend to different curvatures under the shear stresses, and normal stresses will be induced (also near the lateral sides) to maintain curvature compatibility of the plate. When the normal stress is tensile, interlaminar failure is facilitated. The analysis of interlaminar failure is an advanced topic beyond the scope of this book. For the interested reader, a more thorough discussion of composite damage and failure, together with various analytical approaches, can be found in Herakovich (1998).
5.5 Analysis of composite sections 5.5.1 Equivalent elastic properties for the symmetric laminate In this section, we will deal with the analysis of composite sections made of symmetric laminates. The first step is to obtain the equivalent elastic properties. When the composite laminate is symmetric in its ply arrangements, A16 and A26 are both equal to zero. For such a case,
274 Glass FRP for bridge deck replacement
⎡A11 A12 0 ⎤ ⎡Nx ⎤ ⎡A11 A12 0 ⎤ ⎡εox ⎤ [A] = ⎢A21 A22 0 ⎥ and ⎢Ny ⎥ = ⎢A21 A22 0 ⎥ ⎢εoy ⎥ (5.24) 0 A66 ⎦ 0 A66⎦ ⎣γoxy ⎦ ⎣0 ⎣Nxy ⎦ ⎣0 For a laminate of total thickness t, the above can be re-written as:
⎡ εox ⎤ ⎡ a11 a12 0 ⎤ ⎡ Nx ⎤ ⎡ a11t a12t 0 ⎤ ⎡ Nx /t ⎤ ⎢ εoy ⎥ = ⎢ a21 a22 0 ⎥ ⎢ Ny ⎥ = ⎢ a21t a22t 0 ⎥ ⎢ Ny /t ⎥ 0 a66t⎦ ⎣ Nxy /t ⎦ ⎣ γoxy ⎦ ⎣ 0 0 a66⎦ ⎣ Nxy⎦ ⎣ 0 ⎡a11t a12t 0 ⎤ ⎡σx ⎤ = ⎢a21t a22t 0 ⎥ ⎢σy ⎥ 0 a66t ⎦ ⎣τxy ⎦ av ⎣0
(5.25)
The [a] matrix is the inverse of the [A] matrix, with its component given by: a11 = A22A66/det[A] a12 = a21 −A12A66/det[A] a22 = A11A66/det[A] a66 = 1/A66 where det[A] = [A11A22 − (A12)2]A66 is the determinant of the matrix. Assuming the laminate to be a homogeneous material, the constitutive relation can be represented by equivalent moduli and Poisson’s ratios as:
⎡1 ⎤ ⎢ Ex ⎢ εo ⎥ = ⎢ −vxy ⎢ y ⎥ ⎢ Ex ⎣ γoxy ⎦ ⎢ 0 ⎣ ⎡ εoy
−vyx Ey 1 Ey 0
⎤ ⎥ ⎥ 0 ⎥ 1 ⎥ Gxy ⎦ 0
⎡σx ⎤ ⎢σ ⎥ ⎢ y⎥ ⎣τxy ⎦ av
(5.26)
Comparing Eqs (5.25) and (5.26), a direct relation between the equivalent elastic properties and the components of the [a] matrix can be obtained. (Ex)eq = 1/(a11t)
(5.27a)
(Ey)eq = 1/(a22t)
(5.27b)
(Gxy)eq = 1/(a66t)
(5.27c)
(νxy)eq = −a21/a11
(5.27d)
(νyx)eq = −a12/a22
(5.27e)
Similarly, if we define [d] as the inverse matrix of [D], the equivalent bending stiffness (EI)x per unit width for plate bending in the x–z plane is
Glass FRP for bridge deck replacement
275
given by (1/d11), and stiffness (EI)y per unit width in the y–z plane is given by (1/d22). 5.5.2 Stress and deformation analysis of composite members Using the equivalent elastic properties derived above, the stiffness of thinwalled composite sections can be derived for the calculation of deformation under applied loading. Based on the deformations, the force and/or moment per unit width on each part of the member (e.g. a web or a flange) can be obtained. The stresses in the individual plies forming the particular part are then calculated with the approach described in Section 5.4. In the following, the analysis is illustrated with two examples of composites sections, the I-section and the box section. 5.5.2.1 The I-section An I-section under a combination of axial load (N Tx), bending moment (MTx ) and torsion (TTx) is shown in Figure 5.16. Since we will focus on elastic analysis alone, the effects of various kinds of loading can be considered separately and then superposed. Under axial loading, the strain εx is related to N Tx by: N Tx = [2btf Ex,f + htw Ex,w]εx, or N Tx = (EA)eq εx = [2b/(a11)f + h/(a11)w]εx
(5.28)
where the subscripts ()f and ()w represent the web and flange respectively. After finding εx, the force/width on the flange and web, Nx,f and Nx,w, are obtained as: Nx,f = 1/(a11)f εx
(5.29a)
Nx,w = 1/(a11)wεx
(5.29b)
Figure 5.16 Definition of dimensions for the I-section and box-section.
276
Glass FRP for bridge deck replacement
Knowing Nx,f and Nx,w, the stresses in each ply within the flange and web are calculated according to the classical laminate plate theory. Under applied moment MTx, the curvature κx is given by: MTx = (EI)eq κx = {2[b/(d11)f + b(h/2)2/(a11)f] + h3/(12a11)w} κx
(5.30)
In the above expression, the first term involving (d11)f represents the moment of inertia of the flange about its own mid-plane. This is often small compared to the second term which gives the stiffness about the centroid of the I-section, and can hence be ignored in practice. Assuming the plane section remains plane, the strain ε(z) along the xdirection at any distance z from the centroid of the section is given by (κxz). Depending on the location of a particular ply, the lateral strain can be deduced by multiplying (κxz) with the equivalent Poisson’s ratio of the flange or web accordingly. Once the strains in both directions are known, the stress components are calculated by multiplying the strains with the stiffness matrix of the ply. Note that for a flange restrained by a stiff slab above, the lateral strain can be taken to be zero. With torsion TTx applied to the section, the rate of twist β (= dθ/dx) is: TTx = (GJ)eq β = [2bt2f /(3a66)f + ht w2 /(3a66)w]β
(5.31)
The torsion and width acting on the flange and web are then given by: Tf = [t2f /(3a66)f ]β 2 w
Tw = [t /(3a66)w]β
(5.31a) (5.31b)
With Tf and Tw, the stresses in each ply are calculated from classical laminated plate theory. In the above, we have focused on the calculated of stresses. The equivalent stiffness (EA)eq, (EI)eq and (GJ)eq can be used in the calculation of member deflection using conventional approaches. In the calculation of deflection under bending, it should be noted that the ratio of shear to flexural stiffness of composite laminates is usually much smaller than that for homogeneous materials. It is therefore advisable to consider shear deformation in deflection calculations. For the I-section, one can assume all the shear force to be carried by the web, and obtain the shear stiffness (GA)eq as: (GA)eq = h/(a66)w
(5.32)
To calculate the shear deformation, a shear factor k = 1.2 should be applied. For details, the reader should refer to textbooks on mechanics of materials.
Glass FRP for bridge deck replacement
277
5.5.2.2 The box section For the box section, the effects of axial load and bending load are analyzed in exactly the same way as the I-section, using the following equations for the equivalent stiffness: (EA)eq = 2[b/(a11)f + h/(a11)w]
(5.33)
(EI)eq = 2{[b/(d11)f + b(h/2)2/(a11)f] + h3/(12a11)w}
(5.34)
(GA)eq = 2h/(a66)w
(5.35)
Under torsion, the behavior of a closed section (e.g. a box) and an open section (e.g. an I-section) is very different. For the closed section, the shear per unit width (Ns) is uniform along the wall. The following equation then arises from the equilibrium of torsional moment: TTx = (Nsb)h + (Nsh)b = 2(bh)Ns
(5.36)
With Ns, the stresses in each layer within the flange or web can be calculated. The torsional stiffness is given by: (GJ)eq = 2(bh)2/[ b(a66)f + h(a66)w]
(5.37)
As shown in Figure 5.16, the width b of the box is taken to be the total flange width minus the width of the web. This is different from the case of the I-section where b is taken to be the total width of the flange. The reason is due to the uncertainly of fiber arrangements at the corners of the box section, especially when wrapping has been performed as part of the fabrication process. By ignoring the material in the joints, a conservative estimate of the torsional stiffness is obtained. With knowledge of the fabrication process and the actual fiber arrangements, the engineer should be able to make an appropriate judgment on the value of b and h to be used for a particular composite section.
5.6 Summary The replacement of concrete bridge decks with GFRP components is a viable technique to extend the lifetime of bridges. In this chapter, the fabrication method of GFRP bridge deck components first is introduced. Then, a theoretical approach to the analysis and design of laminated fiber composites as well as fiber reinforced plastic structural components is provided. This chapter can hopefully serve as a starting point for civil engineers who are not familiar with composite manufacturing and/or analysis, but are interested in investigating the use of GFRP bridge deck components in real-world projects.
6
Strengthening of reinforced concrete structures with fiber reinforced polymers
6.1 Introduction A significant portion of the world’s civil infrastructure, including highways, bridges, buildings, hydraulic structures and dams, is constructed with reinforced concrete. After years of service, many of these structures are deteriorating. To recover the original factor of safety, strengthening of the structure needs to be performed. Also, for many structures, the current load demand is significantly beyond the value adopted in the original design many years ago. As a notable example, there are 40,000 bridges in the United Kingdom falling short of the new requirements of European highways to carry 44-tonne vehicles. Such bridges have to be strengthened to satisfy current needs. Conventionally, a concrete structure can be strengthened through the enlargement of its members by casting additional concrete, together with the incorporation of additional steel reinforcements. This approach, while simple in principle, has several disadvantages that can make the retrofitting process a real nuisance in practical situations. First of all, to cast new concrete, formwork has to be placed around an existing member and a falsework system set up beneath the formwork is normally required to provide the support. The process will therefore take up significant space under the member to be strengthened. If the member is a beam or a slab in the ceiling of a room, the room has to be vacated for several weeks, to allow sufficient time for the preparation/execution of structural repair, and for the concrete to cure and harden. In other words, the room will be unusable for an extended period of time. To strengthen the girders of a flyover bridge above an underlying highway, one or more lanes of the highway may have to be blocked to accommodate the falsework system. The efficiency of transportation is affected, and this may lead to severe traffic jams during the rush hours. Second, when strengthening needs to be performed in a relatively confined space, transportation of wood for formwork construction, steel members for the falsework system as well as the steel reinforcements and concrete for strengthening can be difficult and labor-intensive. The cost of the process will then be greatly increased. Third, the strengthened member is
Strengthening concrete structures with FRP
279
larger than the original one, implying reduction in headroom and space between members. Fourth, the additional concrete increases the dead-weight of the member, which increases the inertial force acting on the structure when earthquakes strike. Alternatively, concrete structures can be strengthened with the use of steel members. For examples, a steel I-beam can be placed alongside a degraded concrete beam to support the loading from the slab. To strengthen a concrete column, one can either employ steel struts to share some of its loading, or confine it with steel jackets. The use of steel members, however, has its own disadvantages. Similar to the casting of new concrete, the procedures involve in the addition of steel members will take up significant space around the concrete member for an extended period of time. Also, significant weight is added to the structure while space/headroom is reduced. Considering the weight of steel, lifting it into the right position requires the use of powerful machinery. Also, the in-situ processes to ensure the proper connection of the steel member to the existing concrete structure can be rather complicated. All of these will translate into high costs. Moreover, exposed steel members are prone to rusting, so corrosion prevention is an additional issue to be considered. Besides the use of steel structural sections, concrete beams can also be strengthened with the use of bonded or bolted steel plates, on or near the tensile surface (for flexural strengthening) or on the sides (for shear strengthening). Steel plates are lighter than steel I-sections but still very heavy. Powerful machinery is required to move them into place. Also, if adhesive bonding is performed, a temporary supporting system is required, and this will again take up a lot of space. To ensure durability of the retrofit, corrosion protection has to be performed as well. Due to the limitations of conventional strengthening methods, a new approach to strengthen concrete members through the adhesive bonding of fiber reinforced polymers (FRP) was developed in the 1980s and has been applied all over the world. Design recommendations for this strengthening method have been developed by ACI in the United States, FIB in Europe and JSCE in Japan. In the following, the principle of FRP strengthening is first introduced, together with a description of common FRP materials and adhesives. The strengthening procedures are then described. Then, the behavior of concrete beams strengthened in flexure and shear, as well as concrete columns strengthened in compression, will be discussed in detail. For each kind of strengthened member, a design approach will also be introduced.
280
Strengthening concrete structures with FRP
6.2 Structural strengthening with bonded fiber reinforced polymer (FRP) 6.2.1 Materials for FRP fabrication FRP for the strengthening of concrete members is available in different forms, including pre-fabricated plates or shells as well as fiber sheets, fabrics and tows (note: tows are continuous fiber bundles that are well aligned). Pre-fabricated plates are produced by the pultrusion process. Shell elements can be made by hand lay-up or wrapping around a mandrel. Elements formed by wrapping can be cut to form open sections for fitting onto structural members. Alternatively, it is possible to produce a single cut along the longitudinal direction, which allows the shell to “open up” and then be placed around a concrete column. To bond pre-fabricated plates or shells onto concrete members, adhesives are used. When fiber sheets, fabrics, and tows are used, the FRP is often formed directly on the member by applying alternate layers of resin and fiber until the required thickness is obtained. Alternatively, the FRP can be prepared at the site first with the use of fiber and resin. While the resin is still wet, the FRP can be bonded to the concrete member. Some material suppliers pre-impregnate their fiber products with resin. Depending on the product, the use of additional resin during in-situ installation may or may not be necessary. For systems requiring additional resin, the hardener is placed in the additional resin alone, and the FRP hardens at ambient temperature. For pre-impregnated fiber products that do not require additional resin, curing needs to be performed at a higher temperature, so a heating source is required. Three kinds of fiber, made of glass, carbon and aramid, are commonly used in the fabrication of FRP for structural strengthening. Each has its own advantages and limitations. For glass fiber, advantages include its high strength (>2 GPa), high ultimate strain (>3 percent) and relatively low cost. The specific gravity is about 2.6. The major disadvantages include low modulus (E = 72 GPa), sensitivity to abrasion and alkaline environment, as well as low resistance to moisture, sustained loads and cyclic loads. The most commonly used fiber is called E-glass, which is for general purpose. S-glass has better mechanical properties but is more susceptible to alkaline attack. AR-glass (i.e. alkaline-resistant glass) has improved durability in an alkaline environment, but the degradation problem is not completely eliminated. Although glass fibers in FRP are embedded inside a polymer matrix, alkaline attack by ions penetrating the matrix is still a concern. Carbon fibers possess much higher strength and stiffness than glass fibers but are far more expensive. The strength of carbon fibers ranges from 2.2–5 GPa, its modulus from 800–250 GPa, and the ultimate strain from 0.3–1.8 percent. The specific gravity ranges from 2.2 for high modulus fiber to 1.8 for low modulus fiber. It should be noted that the stronger fibers are
Strengthening concrete structures with FRP
281
associated with lower moduli and higher ultimate strain. Besides high strength and stiffness, carbon fiber also possesses excellent resistance to moisture and chemicals, and is insensitive to fatigue. The weaknesses of carbon fiber are its low impact resistance and low ultimate strain. Aramid fibers have very good properties when put in direct tension. The strength ranges from 2.8–4.1 GPa, the modulus from 80–190 GPa and the failure strain can be up to 4 percent. The specific gravity is about 1.4–1.5. However, flexural and compressive properties are relatively poor. Also, aramid fibers creep significantly when exposed to moisture and degrade under UV radiation. The cost of the fiber is in the same range as carbon fibers. For each of the above fibers, the stress–strain relation stays linear up to failure. Final failure of the fiber is due to fracturing with no signs of ductile yielding. With fibers dominating their behavior, common FRP plates and sheets for strengthening also exhibit a linear stress–strain relation and a brittle rupture failure. The relation of this behavior to structural design will be discussed later. Generally speaking, FRP can be prepared with various kinds of thermoplastics and thermosets as matrix material. In practice, all commonly available FRP systems for the strengthening of concrete members employ epoxy as the matrix material. Although more expensive than other thermosetting polymers (such as polyester and vinylester), epoxy has high strength, good creep resistance, strong resistance to chemicals and solvents as well as low shrinkage and volatile emission. For the bonding of prefabricated FRP to a concrete member, epoxy adhesive is also commonly used. The properties of FRP depend on the kind of fiber employed and the percentage of fiber in the composite. Based on values reported in the literature, the typical ranges of mechanical properties for carbon fiber reinforced polymers (CFRP), aramid fiber reinforced polymers (AFRP) and glass fiber reinforced polymers (GFRP), as well as qualitative information regarding long-term performance under various conditions (fatigue, chloride environment, etc.) are listed in Table 6.1. Corresponding properties of steel are also included for comparison. Note that the values in the table are for general reference only. When a certain FRP system is to be employed, specific properties should be obtained from the material supplier. Table 6.1 Typical properties of fiber reinforced polymers and steel
Density (kg/m3) Elastic modulus (GPa) Tensile strength (MPa) Fatigue Sustained loading Alkaline environment Acid/chloride exposure
Carbon FRP
Aramid FRP
Glass FRP
Steel
1600–900 56–300 630–4200 Excellent Very good Excellent Excellent
1050–1250 11–125 230–2700 Good Adequate Good Very good
1600–2000 15–70 500–3000 Adequate Adequate A concern Very good
7800 190–210 250–500 Good Very good Excellent Poor
282
Strengthening concrete structures with FRP
6.2.2 Strengthening principles and configurations The flexural strengthening of concrete beams with bonded FRP is illustrated in Figure 6.1. With loading acting on the beam, the FRP at the bottom is subjected to tension. The tensile force in the FRP will generate an additional moment about the neutral axis of the member to increase the total bending capacity of the beam. To maximize the effectiveness of strengthening, unidirectional fiber sheets, pultruded fiber plates or fabric with fibers running predominantly along the longitudinal direction should be used. The bonded FRP should cover the region that needs to be strengthened, with additional anchorage lengths on both sides to ensure effective stress transfer from the concrete member to the FRP. Under normal conditions, FRP bonding is performed without unloading of the concrete member (e.g. by jacking it to an un-deformed configuration). In other words, the FRP only contributes to carrying the live load of the member, as the concrete member itself is already carrying the full dead load when FRP is applied. To increase the proportion of loading carried by the FRP, the FRP can be pre-stressed before it is bonded to the beam. With uniform pre-stress applied on the plate before bonding, the bonded plate may peel off easily at the ends due to the presence of high stress concentrations. To resolve this problem, one can apply the full pre-stressing to the middle part of the FRP, and gradually reduce the prestressing force towards the plate end. A mechanical system to perform such kind of pre-stressing operation is commercially available. Reinforced concrete beams are always designed to fail in bending, so the inherent shear capacity is higher than the flexural capacity. However, if the beam is strengthened in flexure, shear strengthening may be necessary to maintain the more ductile flexural failure mode. Figure 6.2(a) illustrates the shear strengthening of beams with the use of individual FRP strips that are either prefabricated plates or fiber sheets/fabrics. The function of the strips is similar to that of steel stirrups. Once inclined shear cracks form along the span, the strips act as ties between the concrete on the two sides of each crack. The tensile forces in the strips will then add to the forces in the
Figure 6.1 Flexural strengthening of concrete beam with bonded FRP.
Strengthening concrete structures with FRP
283
Figure 6.2 Shear strengthening of concrete beam with (a) FRP strips; (b) FRP sheet; (c) side-bonded FRP, (d) U-jacketing; (e) full wrapping.
stirrups to increase the total shear force that can be carried. In this case, to maximize effectiveness, the fibers should preferentially be oriented along the length of the strip. In shear strengthening, the use of strips provides greater flexibility in practical design. By varying the strip thickness and/or spacing, different degrees of strengthening can be achieved. However, the procedures for preparing and bonding a large number of individual strips are very labor-intensive. An alternative approach, which applies a single FRP sheet over the whole region to be shear-strengthened, is illustrated in Figure 6.2(b). With this approach, more material is likely to be used, but the labor cost is reduced. Moreover, by covering up the whole region, further penetration of water and other corrosive agents into the structure is prevented. As illustrated in Figures 6.2(c), 6.2(d) and 6.2(e), shear strengthening can be performed with three different configurations, namely side-bonding, Ujacketing and full wrapping. For side bonding, FRP is applied only on the two vertical sides of the member. This is the only possible approach if the beam is connected to a wall underneath, and cutting of a horizontal slot at the beam/wall junction is not allowed. This configuration is the simplest of the three and involves the least labor. However, the strengthening effectiveness is also the lowest since the FRP can debond easily from the
284
Strengthening concrete structures with FRP
beam when shear cracks are formed. (Note: more detailed discussions on this aspect will be provided in a later section.) With the U-jacketing configuration, FRP debonding at the lower side of the vertical beam surface is controlled. Although the upper part of the FRP can still debond, the degree of strengthening is improved over that of the side-bonding configuration. To further enhance the strengthening, mechanical anchors can be applied at the upper part of the vertical strip. When U-jacketing is performed, it is necessary to round off the corners of the beam to a specified radius of curvature, in order to avoid FRP rupture due to high stresses at sharp corners. If a large increase in shear capacity is required, the most effective approach is to perform full wrapping. Vertical slots are cut through the slab on the two sides of the beam, for a strip of FRP sheet to go through. A thin strip can be wrapped around the beam section (which includes part of the slab) several times until the right thickness is reached. Alternatively, a wet strip of the designed thickness can be prepared first and then inserted through the slots. In the latter case, it is necessary to provide sufficient overlapping on the strip to avoid failure at the joint (see Figure 6.2(e)). While full wrapping is very effective, it is highly labor-intensive. Moreover, in many cases, cutting through the slab may not be allowed. When U-jacketing or full wrapping is performed, there is an additional benefit. The bottom strip, which runs in the transverse direction, can resist the separation of the longitudinal FRP from the beam and hence increase the effectiveness of flexural strengthening. Research on the use of inclined strips for shear strengthening has been carried out. With the strips making a higher angle to the crack, the effectiveness is improved. However, the use of angled strips is only possible if there is no load reversal. If the beam is under dynamic loading that can create shear cracks in two inclined directions, the approach is ineffective unless two sets of inclined strips at opposite angles are employed. Compressive strengthening with FRP is based on the enhancement of concrete strength through lateral confinement. When a concrete column is loaded, internal damage leads to an increase in Poisson’s ratio, and a consequent increase in lateral displacement. If a circular column is surrounded by FRP, the tensile stretching of FRP will provide inward pressure on the column. This confining pressure resists the lateral deformation of the concrete and significantly enhances its strength along the vertical direction. The compressive load capacity is therefore improved. For maximum effectiveness, the fibers in the FRP should be running along the hoop direction of the column. Various configurations for column strengthening are illustrated in Figure 6.3. In Figure 6.3(a), the column is wrapped by a single continuous FRP sheet, but individual strips or a continuous helical strip can also be employed. In addition, prefabricated FRP elements in the form of half-shells or shell with a vertical slit can be used. In the former case (Figure 6.3(b)), there should be sufficient lap length between the half-shells to avoid failure at the joint. In the latter case (Figure 6.3(c)), the shell element should be thin
Strengthening concrete structures with FRP
285
Figure 6.3 Strengthening of concrete column with (a) FRP wrapping; (b) prefabricated FRP shell; (c) prefabricated thin FRP shell with a slit.
enough for it to be easily opened at the slit and wrapped around the column. To come up with the required total thickness, several shell elements are required, and the slit in each element should be at different locations around the member circumference. In design, the thickness of an N-layer shell is taken to be (N − 1) times the thickness of each layer. According to the strengthening principle, FRP wrapping is most effective for circular columns, and is applicable to elliptical columns as well. For rectangular columns, the tension in the FRP only produces significant inward pressure at the corners. The strengthening is only effective if the corners are properly rounded. In this chapter, we will only focus on the FRP strengthening of circular columns. The strengthening of rectangular columns will only be briefly discussed. 6.2.3 The strengthening process Various commercial FRP systems are available for the strengthening of concrete structures. Normally, each material supplier will provide clear guidelines related to the use of their system, and these should be closely followed
286
Strengthening concrete structures with FRP
in practice. While the detailed procedure may vary for different systems, the basic principle is the same. In the following, the general strengthening process is presented with explanations of major steps. Before the FRP system is applied to a concrete member, damage or cracks on the member should first be properly repaired (with the approaches discussed in Chapter 4). If there is plaster or other soft material on the concrete surface, it will have to be removed. Bonding of FRP onto a weak substrate will greatly compromise the effectiveness of strengthening as failure would occur within the substrate well before the FRP’s full contribution is attained. After ensuring that the substrate is sound and strong, surface preparation is the next step. Specifically, a thin layer of mortar should be removed from the surface to expose the underlying aggregates. By removing the weak mortar layer and increasing the surface roughness, the FRP/concrete bond can be improved. Common surface preparation methods include blasting with sand, grid or water jet, and surface grinding using mechanical tools. The choice of method depends on the FRP system as well as the kind of strengthening to be performed. For example, blasting is more efficient than grinding but it produces a surface that is more uneven. It is suitable for beams if the FRP system consists of prefabricated strips with significant stiffness and an adhesive that is effective in filling up voids on the surface. However, if dry or uncured fiber sheets that tend to follow the concrete surface profile are employed, the grinding process is preferred. When FRP has to be bent around the edge of a beam (e.g. in U-jacketing or U-wrapping for shear strengthening) or a rectangular column, the corner needs to be rounded to avoid premature failure due to high stress concentration. Theoretically speaking, the higher the radius of curvature, the lower the reduction in FRP strength due to concentrated stress. In practice, the radius of curvature at any corner should be 25 mm or above. In the strengthening of columns, the concrete surface should be kept convex everywhere for the member to be effectively confined at all locations. In this situation, excessive unevenness on the surface is also undesirable. Even when grinding is performed, the prepared surface may still be overly uneven. In this case, a sticky polymeric paste that can bond well with concrete, called a putty, can be applied to smoothen the surface first. For most commercial FRP strengthening systems, a putty compatible with the adhesive or resin is available from the supplier. After the concrete substrate is repaired (if necessary) and the surface properly prepared, the FRP can be applied. Before the bonding of FRP, dust, dirt, oil and other contaminations should be completely removed from the concrete surface. In addition, the surface needs to be dry. The adhesive or resin, which is normally supplied in two parts, can then be mixed according to the proportions recommended by the supplier. For prefabricated plates, adhesive is applied on both the concrete surface and one side of the plate. For the adhesive on the plate, it should be applied to form a dome-shaped cross-section with the middle part slightly higher than the side. The plate is
Strengthening concrete structures with FRP
287
then lifted to the right position and pressed onto the concrete surface with a rubber roller. With the adhesive slightly thicker in the middle, the lateral movement of adhesive during the bonding process will push air to the side and significantly reduce the probability of air being trapped within the adhesive. When fiber sheets or fabrics are employed with resin, a primer is sometimes applied first. The primer is usually of lower viscosity than the resin, and penetrates better into the concrete substrate. Then, the first layer of resin is applied, followed by a layer of fiber sheet, and an additional layer of resin. To facilitate the impregnation of resin into the FRP sheet, and to remove trapped air, a roller brush is used to apply pressure on the layers of resin and fiber. After one layer is completed, the resin is left to cure for a short period of time before the second layer is applied. This is to prevent the movement of the first layer of material during the applying of the next layer. The procedure is repeated until the required FRP thickness is reached. For the repair of columns, a semi-automatic wrapping process has been developed. The concept is illustrated in Figure 6.4. A ring holding fiber spools is first placed around a concrete column. Fiber tows from the various spools are then glued to a number of points around the circumference of a section near the top or bottom of the column. By rotating the ring and moving it vertically at the same time, the fibers are wrapped around the column surface, with the fiber angle depending on the speeds of rotation and vertical movement. In this process, the fiber tows can be passed through a resin bath before they are wrapped around the column. The composite can then be cured at ambient temperature. Alternatively, pre-impregnated tows
Figure 6.4 Schematics of an automatic FRP wrapping system.
288
Strengthening concrete structures with FRP
can be used, but then curing needs to be performed with the use of a heating source such as a cylindrical heater around the column. Due to the light weight of fiber sheets/fabrics and FRP plates, the epoxy resin or adhesive is sufficient to hold them in place under most situations, so temporary supporting systems are not required. In some cases, a simple clamping system providing small compression on the FRP is sufficient to prevent its movement before the epoxy hardens. Epoxy for FRP strengthening systems can undergo curing at room temperature, and gain sufficient strength within a week. After curing, the strength of epoxy is higher than that of the underlying concrete. Providing the surface has been properly prepared, failure of the bond will occur within the concrete substrate rather than along the epoxy/concrete interface. 6.2.4 Advantages and limitations In comparison to conventional strengthening methods, the major advantages of FRP strengthening arise from the high strength/weight ratio of FRP and the relative simplicity of the retrofitting process. Take, for example, a typical CFRP with strength of 4200 MPa and relative density of 1.8. Considering the possibility of pre-mature debonding failure and variability of the FRP bonding process, we may assume the CFRP stress at structural failure to be about 2000 MPa. To carry the same tensile force as a steel plate with a yield strength of 500 MPa and relative density of 7.8, the weight of the equivalent CFRP plate is only 6 percent that of the steel member. A significant reduction in material weight can also be shown for other kinds of FRP. In most cases, the weight of the FRP element (plate or shell) is low enough for it to be handled by between one and three workers without the need of heavy machinery. If a large and heavier element is required for strengthening, it can always be produced directly on the concrete member with thin FRP sheets. Using uncured individual sheets, the shape of concrete members (such as the round surface of a cylindrical column) can easily be followed. Bonding is also a simple and efficient process that can be performed during the off-hours of facilities, such as the time between midnight and the early morning. The disturbance to users of the facilities is hence minimal. After the FRP is bonded, it does not require temporary support. In common strengthening applications, the required thickness of FRP seldom exceeds several millimeters. The reduction in headroom/space and increase in weight of structure are both negligible. The FRP bonding technique, while being efficient and effective, is not without its limitations. Fire resistance is always a concern. At high temperature, the load-carrying capacity of FRP is greatly reduced by the softening of resin or adhesive. It is necessary to ensure that the loss of FRP strengthening will not lead to sudden collapse. According to the ACI recommendations (2002), the non-strengthened structure should be able to carry the unfactored dead and live loads. This seems to be a reasonable requirement that
Strengthening concrete structures with FRP
289
can be adopted in general. Also, as polymeric materials can actually burn, any adhesive or resin used in the FRP should not increase the severity of the fire. In some cases, the load-carrying capacity of the FRP has to be maintained in a fire. An example is a bridge that has been widened, so the columns need to be strengthened to carry additional dead and live loads. Under this situation, a fire protection system should be installed. An external polymer layer that can foam up during heating to insulate the FRP is effective in protecting the FRP from fire. Another common concern about the use of FRP is the possibility of attack by ultraviolet (UV) light from the sun. This will be a problem for FRP around an exposed column or FRP on the lateral sides of an exposed beam. UV light tends to break down molecular chains in a polymer, and hence will degrade the mechanical performance of the polymer and the FRP. With temperature change and moisture penetration, the effect may become more severe. To protect the FRP against UV light, the simplest way is to apply a light-colored paint that will reflect most of the energy away. While FRP can easily be bonded to a concrete structure, it can also easily be removed. As a result, FRP strengthened structures are vulnerable to vandalism. For critical locations, access to strengthened parts of the structure may have to be restricted. As FRP bonding is a new technique that is still under research and development, the lack of a universally accepted approach for failure analysis limits its practical application. While various design guidelines have been developed separately in the USA, Europe, Japan and other countries, the design equations in various guidelines may be very different. As a result, the predicted load capacity of a strengthened member may vary significantly among different guidelines. With continuous research and development activities on FRP strengthening of concrete structures, the situation is likely to improve in the coming years. A widely accepted approach for the analysis and design of strengthened member will eventually evolve and lead to the convergence of various guidelines. In the following sections, the mechanics of FRP strengthening will be discussed separately for beams under flexure and shear, and columns under compression. The focus of discussion is on failure mechanisms and factors affecting the failure load. For each kind of FRP strengthening, a design framework is presented together with design equations to predict the contribution of FRP to the load-carrying capacity. While these equations are likely to be modified or replaced in the future, the overall framework should still be applicable for practical design.
290
Strengthening concrete structures with FRP
6.3 Flexural strengthening of beams 6.3.1 Failure modes For a FRP-strengthened beam under flexural loading, failure can occur in many different modes. Generally speaking, one can distinguish between the situation with full composite action and that with a loss of composite action. Full composite action takes place when there is perfect bonding between the FRP and concrete, or limited debonding that has little effect on the FRP force. However, in many cases, loading results in significant debonding, so the FRP is no longer effectively coupled to the concrete member. In other words, the composite action is lost so the FRP force and the level of strengthening are greatly reduced. In the following, the various failure modes are discussed. 6.3.1.1 Failure modes under full composite action When there is full composite action, the failure modes of strengthened beams are similar to those for normal reinforced concrete members, with the additional mechanism of FRP rupture when its ultimate strength is reached. Plausible failure modes include: (1) steel yielding followed by concrete crushing; (2) steel yielding followed by FRP rupture; (3) concrete crushing before steel yielding; and (4) shear failure. The occurrence of mode (1), (2) or (3) depends on the area fraction of steel and FRP at the critical crosssection. Before strengthening, a properly designed beam should fail with steel yielding followed by concrete crushing. If only a small amount of FRP is added, the same failure mode (Mode (1)) should be maintained. However, if a large volume of FRP is bonded to the bottom of the member, the neutral axis will exhibit a significant downward shift that increases the concrete strain at the top and reduces the strain in the tensile steel. Crushing of concrete (Mode (3)) may hence occur before steel yielding, which is undesirable. Also, depending on the failure strain of the FRP, it may rupture after steel yielding before the concrete crushes (Mode (2)). Theoretically speaking, FRP may also rupture before steel yielding. If this is the case, the strengthening is not effective at all, because the contribution from the steel reinforcements is not yet fully activated when the FRP fails. For properly designed FRP strengthened beams, this mode of failure should be eliminated. Mode (4) occurs when the increase in load capacity due to flexural strengthening exceeds the original shear capacity of the member. In this case, strengthening in shear is also required, and this will be discussed in Section 6.4. 6.3.1.2 Failure modes with loss of composite action Loss of composite action is caused by stress concentrations along the FRP/ concrete interface. The locations along the beam with stress concentrations
Strengthening concrete structures with FRP
291
are illustrated in Figure 6.5. The FRP stress is zero at the section where the plate is terminated (or cut off). For tensile force to develop in the FRP plate, interfacial shear stress must exist near the plate end for equilibrium to be satisfied. Detailed analysis, first performed by Roberts (1989), reveals the presence of a concentrated shear stress at the cut-off point. Also, to have compatible curvatures for the concrete beam and FRP plate, concentrated tensile stress normal to the interface is also generated. Under shear and tensile stress concentrations, local failure initiates at the plate end with the formation of an inclined crack that propagates upwards (Figure 6.6(a)). Once the crack grows beyond the steel reinforcements, load transfer to the concrete cover will occur mainly through shear and normal stresses at the level of the reinforcements (Figure 6.6(b)). Under increased loading, a horizontal crack forms at the level of the steel reinforcement and its propagation will cause the concrete cover near the plate end to debond together with the FRP. In the literature, this failure mode is referred to as “plate end debonding”, “cover separation” or “peel-off”. It should be pointed out that initial cracking at the cut-off point can occur at a load level significantly below the ultimate load of the strengthened member, even if the plate is terminated close to the support (as in the specimen shown in Figure 6.6(a)). As a result, while the interfacial shear stress concentrations are important for the initiation of debonding at the plate end, they are not applicable to the prediction of ultimate failure load (Leung 2007). Stress concentration is also present at the bottom of flexural or flexural/ shear cracks along the span of the concrete beam. Under loading, these cracks tend to open but their openings are restrained by the FRP plate. In return, concentrated shear stresses are induced along the interface on the two sides of the opening crack (Leung 2001). Under this situation, debonding initiated at the vicinity of the concrete crack, and propagates towards the end of the plate. The mode of failure is referred to as crack-induced debonding. For this failure mode, material separation occurs within the concrete at a small distance from the concrete/adhesive interface. This
Figure 6.5 Locations with stress concentrations along the FRP/concrete interface.
292
Strengthening concrete structures with FRP
Figure 6.6 Debonding failure at the plate end.
is evidenced from the thin layer of concrete (of about several mms) attached to the FRP plate after debonding failure occurs (Figure 6.7(a)). The occurrence of debonding within the concrete member at a distance from the concrete/adhesive interface can be explained as follows. First, penetration of adhesive into the concrete may increase the strength of a thin layer of material right next to the interface. Second, high shear stresses acting along the concrete/adhesive interface will produce micro-cracks that tend to propagate away from the interface at a certain angle along the principal compression direction (Figure 6.7(b)). The interaction and coalescence of these inclined cracks will produce the final debonding surface inside the concrete substrate. 6.3.2 Factors affecting flexural load capacity when there is loss of composite action The flexural load capacity of FRP strengthened beams can be obtained with the same approach for conventional reinforced concrete, as long as the FRP force at failure is known. When there is full composite action, the FRP
Strengthening concrete structures with FRP
293
Figure 6.7 Crack-induced FRP debonding.
behaves as a linear elastic material up to its rupture strength (or, till concrete crushing occurs), and its failure stress is easy to find. When plate-end debonding or crack-induced debonding occurs, various models have been proposed to obtain the ultimate failure load directly, or to calculate the FRP stain at the instant of failure. Unfortunately, a universally accepted model is yet to be developed, so different approaches for calculating the debonding failure load are adopted in different design recommendations. Detailed discussion of the existing models is beyond the scope of this book. The interested reader should refer to Teng et al. (2002), where an excellent summary of various modeling approaches can be found. In what follows, we will first discuss the effect of various parameters on the debonding failure load. Then, an empirical formula for predicting the FRP failure strain, based on the fitting of a large number of test data, is presented. To facilitate the discussions, the following terms are defined (see also Figure 6.8): S h d b
= shear span of the beam = total depth of the beam = effective depth of the tensile reinforcement = width of the beam
294
Strengthening concrete structures with FRP
Figure 6.8 Definition of geometrical parameters for the strengthened beam.
bp = width of the FRP plate tp = thickness of the FRP plate a = cut-off distance, which is the distance between the plate end and the closer support Ep= Young’s modulus of FRP fc′ = compressive cylinder strength of concrete. 6.3.2.1 Effect of cut-off distance The variation of FRP debonding load with cut-off distance is illustrated by a set of test data given in Table 6.2. When the cut-off distance is sufficiently small (i.e. the plate end is sufficiently close to the support), the failure load and maximum FRP strain are essentially independent of cut-off distance. However, as the cut-off distance increases, significant reduction in strengthening effectiveness can be observed, with the failure mode changing from crack-induced debonding to plate end failure. The trend of experimental results can be explained with the help of Figure 6.9, which shows schematically the variation of crack-induced debonding load and plate end failure load with cut-off distance. For crack-induced debonding, the maximum FRP force depends on the distance of the plate end from the major crack initiating the debonding process. As the length of plate beyond the major crack increases (i.e. the cut-off distance decreases), both experimental bond tests and theoretical models indicate that the FRP force would approach an Table 6.2 Effect of cut-off distance on strengthening effectiveness Cut-off distance a (mm)
a/S
Failure Load (kN)
Max. FRP strain at failure (×10−6)
Failure mode
25 300 450 600
0.004167 0.25 0.375 0.50
263.4 265.1 234.8 204.1
7057 7280 4943 3072
Crack-induced Crack-induced End Failure End Failure
Notes: For all beams, b = bp = 150 mm, h = 400 mm, d = 370 mm, S = 1.2 m, tp = 0.44 mm, Ep = 235 GPa and fc′ = 33 MPa. Failure Load of Control Beam (with no FRP) is 197.8 kN. Tests are performed at the Hong Kong University of Science and Technology.
Strengthening concrete structures with FRP
295
Figure 6.9 Transition of failure mode with increasing cut-off distance.
asymptotic value. As a result, the failure load also approaches a constant value as shown in Figure 6.9. For plate end failure, the failure load decreases continuously with cut-off distance, due to increasing stresses in the concrete adjacent to the plate end. Since the failure loads for the two modes do not decrease at the same rate with the cut-off distance, a transition of failure mode may occur. 6.3.2.2 Effect of FRP stiffness Table 6.3 shows the test results on similar beams strengthened with the same FRP of different thickness. In these tests, the width of FRP is kept constant. The stiffness of the FRP, which is the product of its Young’s modulus and area, can hence be represented by Ep tp (as in the ACI Design recommendation). According to the results in Table 6.3, increasing FRP stiffness at Table 6.3 Effect of plate stiffness on strengthening effectiveness Plate thickness tp (mm)
Eptp (×103 N/mm)
Failure load (kN)
Max. FRP strain at failure (×10−6)
Failure mode
0.22 0.44 0.66 0.88 1.10 1.32
51.7 103.4 155.1 206.8 258.5 310.2
216.2 239.1 255.2 275.9 276.8 251.8
7254 6475 5655 4934 4421 3097
Crack-induced Crack-induced Crack-induced Crack-induced End failure End failure
Notes: For all beams, b = bp = 150 mm, h = 400 mm, d = 370 mm, S = 1.2 m, a = 15 mm, Ep = 235 GPa and fc′ = 29 MPa. Failure Load of Control Beam (with no FRP) is 180.8 kN. Tests are performed at the Hong Kong University of Science and Technology.
296
Strengthening concrete structures with FRP
constant width initially increases the failure load and decreases the maximum FRP strain, but eventually leads to reduction in both. The reduction in strengthening effectiveness is associated with a change of failure mode from crack-induced debonding to plate end failure. Based on fracture mechanics arguments, the FRP force for crack-induced debonding to occur should increase with plate thickness. However, as the plate becomes stiffer, the plate end stress concentration also becomes more severe. Once the inclined crack initiated at the plate end propagates beyond the steel reinforcements, higher tensile stresses will be induced at the level of the steel reinforcement by a stiffer cover (which consists of both the concrete and the FRP plate). Cover separation is then easier to take place. Since the debonding failure loads for the two modes vary in opposite trends with the plate stiffness, a transition can be expected. A very interesting observation here is the existence of an optimal FRP thickness (or stiffness) for flexural strengthening. Excessive increase in thickness will lead to a transition to plate end failure and associated reduction in ultimate load. It should be noted that the cut-off distance for the tested beams is very small. According to the discussions in the former sub-section, when the cut-off distance is larger, plate end failure is easier to occur, so the optimal FRP thickness is expected to decrease. From Table 6.3, the maximum FRP strain decreases continuously with plate stiffness. The rate of decrease is greatly increased after the transition of failure mode. When crack-induced debonding occurs, a four-fold increase in stiffness (as tp increases from 0.22 mm to 0.88 mm) leads to about one-third reduction in maximum strain. If plate end failure occurs, 50 percent increase in plate stiffness (due to increase of tp from 0.88 mm to 1.32 mm) results in a reduction of nearly 40 percent. This result raises an issue for practical applications. Real structural members are much larger in size than laboratory specimens. To provide sufficient force for strengthening, thick FRP plates are required, and the significant reduction in strengthening effectiveness is a matter of concern. To address this issue, the effect of member size on flexural strengthening should be studied. 6.3.2.3 Effect of member size Table 6.4 shows two sets of tests on the effect of member size on failure load and maximum failure strain. In each series, the geometry of members and steel reinforcement ratios are kept the same. From the results, two interesting observations can be made. First, while the results for beams with d = 400 mm (Table 6.3) indicate failure mode transition when tp is between 0.88 mm and 1.1 mm, the failure mode for beams with d = 800 mm remains the same for tp up to 1.76 mm. Second, comparing the maximum FRP strain for the same tp but different member size d, the FRP plate is able to carry a higher strain when it is bonded to a larger beam. These observations are encouraging as they show that the reduction in strengthening effectiveness
Strengthening concrete structures with FRP
297
Table 6.4 Effect of member size on strengthening effectiveness Beam tp depth (mm) h (mm)
Cut-off Distance a (mm)
fc′ (MPa)
Load Max. FRP capacity strain at (kN) failure (×10−6)
Failure mode
200 400 800 200 400 800
200 400 800 15 30 60
47 47 47 29 29 29
75.7 273.5 1025.3 74.4 275.9 1097
Crack-induced Crack-induced Crack-induced Crack-induced Crack-induced Crack-induced
0.22 0.44 0.88 0.44 0.88 1.76
9737 7364 6548 5904 4934 4816
Notes: For all beams, b = bp = 0.375 d, S = 3 h, Ep = 235 GPa. Tests are performed at the Hong Kong University of Science and Technology.
for thick plates on large members employed in practice is not as drastic as test results on laboratory specimens may indicate. Summarizing the results in Tables 6.1 to 6.3, one may postulate that the failure mode transition is dependent on a/S and Ep tp /d. The second parameter considers the effect of member size as well as plate stiffness on the failure mode. Either the total depth (h) or effective depth (d) can be used as the dimension parameter. As most equations for conventional concrete design involved d rather than h, d is also employed here. In Figure 6.10, the reported failure modes for 111 tests are summarized in a plot with a/S and
Figure 6.10 Effect of tpEp/d and a/S on the failure mode.
298
Strengthening concrete structures with FRP
Ep tp /d being the two axes. For most of the tests, crack-induced debonding occurs when both a/S and Ep tp /d are relatively small. When one or both of them becomes large, plate end failure occurs. The dotted curve in the figure seems to provide a reasonable boundary between the two failure modes. While Ep tp /d affects the debonding failure mode, members with the same Ep tp /d but different sizes (d) do not exhibit the same maximum FRP strain. This is illustrated by the test results in Table 6.3 for geometrically similar members of different sizes, strengthened with the same kind of FRP. In the ACI design recommendation (2001), an empirical equation for the maximum FRP strain is proposed in terms of the parameter Ep tp alone. However, for a more accurate model, both Ep tp /d and d should be employed as independent parameters.
6.3.2.4 Effect of FRP width ratio Table 6.5 shows the result on beams strengthened with FRP of the same thickness but different width to study the effect of the parameter bp /b. With reduced bp /b, the maximum FRP strain increases to compensate for the reduction in FRP area. When the FRP width is 150 mm, the load of the strengthened beam is 33 percent beyond the control member. With the FRP width reduced to 50 mm (i.e. the total FRP area is reduced to one-third), 17 percent increase in load can still be achieved. The enhanced effectiveness with reduced bp /b is attributed to the presence of concrete material beyond the two sides of the bonded plate, which helps to resist debonding failure. In the flexural design of strengthened beams, this factor should not be overlooked.
6.3.2.5 Effect of concrete cover thickness The concrete cover thickness is related to both the total beam depth (h) and the effective depth (d). According to Raoof and Hassanen (2000), a higher cover thickness increases the likelihood of failure at the plate end. This can Table 6.5 Effect of plate width on strengthening effectiveness Plate width bp (mm)
bp/b
Failure load (kN)
Max. FRP strain at failure
Failure mode
150 100 75 50
1.0 0.667 0.50 0.333
263.4 247.9 239.8 232.2
7057 8401 9475 11441
Crack-induced Crack-induced Crack-induced Crack-induced
Notes: For all beams, b = 150 mm, h = 400 mm, d = 370 mm, S = 1.2 m, a = 25 mm, Ep = 235 GPa and fc′ = 33 MPa. Failure Load of Control Beam (with no FRP) is 197.8 kN Tests are performed at the Hong Kong University of Science and Technology.
Strengthening concrete structures with FRP
299
be explained by the higher stiffness of a thicker cover, which induces higher stresses at the level of the steel reinforcement, making it easier for cover separation to occur. 6.3.2.6 Effect of shear span to effective beam depth As pointed out by Sebastian (2003), FRP debonding is sensitive to the ratio between maximum moment and maximum shear (Mmax/Vmax). When (Mmax /Vmax) increases, the failure mode may go from plate end debonding to crack-induced debonding. For beams tested under four-point loading, which is the most common configuration adopted in the literature, the effect of (Mmax /Vmax) can be represented by the factor (S/d). 6.3.2.7 Effect of concrete properties Debonding always occurs within the concrete, at a small distance from the interface for crack-induced debonding, and at the level of the steel reinforcement for plate end failure. In the literature, the debonding load has been expressed in terms of various material parameters of concrete, including its compressive strength, tensile strength, surface pull-off strength and interfacial fracture energy. Of all these parameters, the concrete compressive strength is the only one consistently reported in experimental studies. Equations to obtain the other parameters in terms of the compressive strength have also been proposed. While the compressive strength may not be the parameter governing the actual failure, it is perhaps the only one that can be employed for developing an empirical model for design. 6.3.3 An empirical model for determining the maximum FRP strain at failure Summarizing the above discussions, the maximum FRP strain at debonding failure can be related to the following parameters: (1) Ep tp /d, (2) d, (3), bp /b, (4) h/d, (5) S/d, (6) a/S and (7) fc′. It should be noted that the adhesive thickness and stiffness have not been included, due to controversial findings regarding the effect of the adhesive on FRP debonding. While some researchers find it beneficial to reduce the shear stiffness of the adhesive layer, others discover no effect. Also, the adhesive thickness is hard to measure and seldom reported. Data related to the adhesive are hence insufficient for the development of an empirical model. In Leung (2006), an empirical approach is proposed to determine the ultimate FRP strain from the seven parameters presented above. With a comprehensive experimental database of 143 tests, a neural network relating the ultimate FRP strain to the various parameters is trained and validated. Using the validated network, an empirical curve for the maximum FRP strain together with several correction equations are generated to
300
Strengthening concrete structures with FRP
provide a simple means to obtain the FRP debonding strain for practical design. For details, the readers should refer to the above paper. Here, only the final results are shown. The design curve given in Figure 6.11 is for the determination of (εpu)ref in terms of Ep tp /d. The maximum FRP strain (εpu) is then obtained from: εpu = (εpu)ref * Cbp/b * Cd * Ch/d * CS/d * Ca/S * Cfc′ /RF
(6.1a)
RF = reduction factor with recommended value of 1.6: Cbp/b = −0.274 ((bp /b)/0.8 − 1) + 1
(6.1b)
Cd = 0.0051 (d/200 − 1) − 0.1453 (d/200 − 1) + 1
(6.1c)
Ch/d = 0.987 ((h/d)/1.2 − 1) + 1
(6.1d)
CS/d = 0.1967 ((S/d)/4 − 1) + 1
(6.1e)
Ca/S = 0.0008 ((a/S)/0.1 − 1)2 − 0.0627 ((a/S)/0.1 − 1) + 1
(6.1f)
2
′ c
Cfc′ = 0.2862 (f /40 − 1) + 1 = 0.2862 (fcu/50 − 1) + 1
(6.1g)
When bp /b = 0.8, d = 200 mm, h/d = 1.2, S/d = 4, a/S = 0.1, fc′ = 40 MPa, all the correction factors become 1.0, and the maximum FRP strain is given by (εpu)ref /RF. (εpu)ref can hence be interpreted as the maximum FRP strain when all the other parameters are at specific reference values. The Ci’s are then correction factors for the general case when the various parameters deviate from their reference values. In Eq (6.1g), fc′ is the cylinder strength of
Figure 6.11 Graph for the determination of (εpu)ref in terms of Ep tp /d.
Strengthening concrete structures with FRP
301
concrete while fcu is the cube strength which can be taken as 25 percent higher, according to the British code. In Figure 6.12 the normalized ultimate moments computed with εpu from Eq. 6.1 is compared with experimental results for 143 tests. The data points for RF = 1.0 (shown as crosses) are obtained from the original expressions generated from the neural network. As one can see, they spread around the 45 degree line within a narrow band, showing good fitting of data with the model. In design, we recommend to employ a reduction factor of 1.6. In this case, less than 5 percent of the predicted values will go below the experimental results. Together with the use of additional material safety factors, this will provide a safe design for practical applications.
6.4 Design of beams strengthened in flexure 6.4.1 General assumptions and material behavior The flexural design of strengthened beams follows closely the conventional method for reinforced concrete design. Specifically, concrete is taken to carry compression only. The “plane section remains plane” assumption is followed and relative slip between FRP and concrete surface is neglected. The latter is justified as the relative slip remains small until FRP debonding becomes unstable at the ultimate load. In the analysis, the following material behaviors for concrete, steel and FRP are assumed.
Figure 6.12 Comparison of normalized ultimate moment obtained from empirical approach and experimental results.
302
Strengthening concrete structures with FRP
6.4.1.1 Concrete The compressive stress–strain behavior for concrete from BS8110 is employed. It is given by the following form: σc = (Ecεc −
Ec 2 ε c) / γm for εc < ε0 2ε0
σc = 0.67 fcu / γm for ε0 ≤ εc ≤ εcu
(6.2a) (6.2b)
where σc and εc are the stress and strain respectively, fcu (N/mm2) is the cube compressive strength, Ec (= 5500冪fcu N/mm2) is the initial tangent modulus of concrete, γm (= 1.5) is the partial safety factor for the strength of concrete, ε0 (= 0.00024冪fcu) is the strain at the end of the parabolic part of the stress–strain diagram and εcu (= 0.0035) is the ultimate strain of concrete. 6.4.1.2 Reinforcing steel Elastic perfectly plastic behavior is assumed for steel reinforcements in both tension and compression. The stress (σs) and strain (εs) are related by: σs = Esεs / γm for εs < εy
(6.3a)
σs = fy / γm for εs ≥ εy
(6.3b)
where Es is the steel elastic modulus, fy and εy are respectively the yield strength and yield strain, and γm (= 1.05) is the partial factor of safety for steel. 6.4.1.3 FRP plate The FRP is taken to be linear elastic until failure occurs either by rupture or debonding. Since failure is brittle, the stress drops to zero right after the maximum value is reached. The FRP stress (ff) is then related to its strain (εf) by: ff = Ef εf / γm for εf < εfe
(6.4)
where Ef and εfe are respectively the Young’s modulus and maximum sustainable strain of the FRP, and γm, the partial factor of safety. Depending on the type of FRP, the kind of application and the failure mode, different values of γm, are given in Table 6.6. The definition of application type in Table 6.6 follows that in the FIB Report (2001). Application Type A refers to
Strengthening concrete structures with FRP
303
Table 6.6 Partial safety factor for the FRP Application type A
Application type B
Type of FRP
Fiber rupture
Debonding
Fiber rupture
Debonding
CFRP AFRP GFRP
1.20 1.25 1.30
1.25 1.30 1.40
1.35 1.45 1.50
1.40 1.50 1.60
pre-cured systems under normal quality control conditions, or wet lay-up systems under conditions with high degree of quality control on both the application conditions and application procedures. Application Type B refers to wet lay-up systems under normal quality control conditions, or any system under difficult on-site working conditions. The safety factors for fiber rupture are also taken from FIB. For debonding failure, we recommend using slightly higher values as the failure load is likely to exhibit higher variability. It should be pointed out that the application of FRP in strengthening still has a relatively short history, so the proposed safety factors are only tentative. Modified values may be used to account for different environmental exposures (especially when GFRP is used). With the accumulation of additional test data and field experience in the future, the proposed factors will be refined. 6.4.2 Initial situation of the member In most cases, the bonding of FRP to a concrete member is conducted without unloading of the concrete structure or pre-stressing of the FRP. The FRP will therefore only support additional loading beyond that when it is applied. To find the strain (and stress) in the FRP, it is important to know the initial strain at the concrete substrate (εfo) during the bonding of FRP. When the concrete strain at the substrate increases to εc (> εfo) on further loading, the FRP strain is given by (εc − εfo). Let M0 be the moment acting on the critical section of the concrete beam during FRP bonding. As M0 is typically larger than the cracking moment Mcr, a cracked concrete section should be employed in the calculation. If M0 is smaller than Mcr, the initial strain εfo is very small and its influence on the FRP strain can be neglected. Considering the cracked section in Figure 6.13 and assuming linear elastic behavior of concrete in compression, the neutral axis depth (xo) is obtained from: 1 2
bx 20 +
Es
冢E − 1冣 A′ (x s
c
0
− d′) =
Es As(d − x0) Ec
(6.5)
304
Strengthening concrete structures with FRP
Figure 6.13 Initial condition of the beam.
where As and A′s are the areas of tensile and compressive steel respectively, d and d′ the corresponding distances of the tensile and compressive reinforcements from the top of the beam, and b is the beam width. The concrete strain (εfo) at the bottom of the concrete beam is then εf 0 =
M0(h − x0) EcIc
(6.6)
where Ic is the moment inertia of the transformed cracked section, given by: Ic =
bx 30 Es Es + − 1 A′s (x0 − d′)2 + As(d − x0)2 3 Ec Ec
冢
冣
(6.7)
On further loading after the FRP is bonded, the stress in the composite plate (ff) is calculated from: ff = Ef εf = Ef (εc − εfo)
(6.8)
where εc is the strain at the tensile surface of the concrete beam and εf is the actual strain in the FRP plate. Theoretically speaking, the change in strain at the level of the FRP is different to that at the concrete substrate. However, as both the FRP and adhesive are very thin, the difference can be neglected in all practical calculations. 6.4.3 A design framework for the strengthened beam A general design framework should cover both the ultimate state and serviceability state. To find the ultimate moment of the section, a systematic
Strengthening concrete structures with FRP
305
approach proposed by Malek and Patel (2002) is followed. Equations will be derived for the rectangular beam. For the analysis of an I-beam (which includes part of the slab above the beam), the interested reader should refer to the original reference. For serviceability considerations as well as avoidance of creep rupture, approaches suggested by ACI (2002) are adopted. 6.4.3.1 Design for ultimate state DETERMINATION OF FRP STRAIN AT FAILURE
Theoretically speaking, if the FRP fails by rupture, the failure strain should be equal to εfu, the failure strain measured in a direct tension test. However, many test results indicate the occurrence of FRP rupture in strengthened beams at a lower strain value. For design, we propose to take the rupture strain to be 0.8εfu. Debonding failure will occur when the strain reaches εpu given by Eq. (6.1). The maximum sustainable strain of the FRP (εfe) is given by the smaller value between 0.8εfu and εpu, divided by the appropriate material safety factor. (Note: as the material safety factors for debonding and rupture are different, 0.8εfu and εpu should be directly compared first. The smaller value is then divided by γm for the corresponding failure mode.) 6.4.3.2 The balanced plate ratio for simultaneous steel yielding and concrete crushing In conventional reinforced concrete design, steel yielding should occur before the crushing of compressive concrete. For the strengthened beam, the same should be ensured. The plate ratio ρf is given by Af /bd, where Af is the cross-sectional area of the composite plate, b the width of the section and d the effective depth to the steel reinforcement. The balanced plate ratio (ρf,b) is defined as the plate ratio at which steel yielding occurs simultaneously with concrete crushing at the ultimate state. Its determination is illustrated in Figure 6.14, where the concrete stress is represented by the equivalent stress block given by BS8110. For the stress block in Figure 6.14, the compressive stress of 0.45fcu has already incorporated a material safety factor of 1.5. In the balanced condition, the tensile steel reinforcement attains its yield strain (εy) while the concrete at the compressive surface reaches its ultimate strain (εcu). With the strain varying linearly along the depth of the beam, the neutral axis position is given by: xb =
εcu εcu + εy
d
(6.9)
From force equilibrium, the balanced plate ratio can be calculated from:
306
Strengthening concrete structures with FRP
Figure 6.14 Determination of balanced plate ratio for simultaneous steel yielding and concrete crushing.
ρf,b =
0.405fcu xb / d + 0.95(ρ′ε′s Es − ρfy) Ef εf
where εf = εcu
(6.10)
h − xb − εf 0 ≤ εfe, ρ = As /bd, ρ′ = A′s /bd xb
In design, ρf should be less than ρf,b to ensure the yielding of steel before crushing of concrete. If more ductility is required, ρf should be limited to a fraction of ρf,b. In Malek and Patel (2002), a fraction of ¾ is suggested. This can be taken as a general guideline, but the specific value should be determined according to experience and specific ductility requirement of the retrofitted member. 6.4.3.3 The balanced steel ratio for simultaneous plate failure and concrete crushing After ensuring the occurrence of steel yielding before failure, the next step is to determine the ultimate failure mode, which can be either FRP rupture/ debonding or concrete crushing. One can define another balanced plate ratio ρf,bb for FRP rupture/debonding and concrete crushing to occur simultaneously. For ρf> ρf,bb, compressive crushing will be the failure mode. Conversely, FRP rupture/debonding will occur. This balanced condition is illustrated in Figure 6.15, with strain in the composite plate reaching εfe and the strain at the compressive surface of the beam reaching εcu. With linear strain variation over the depth of the member, the position of the neutral axis (xbb) is given by:
Strengthening concrete structures with FRP
307
Figure 6.15 Determination of balanced plate ratio for simultaneous FRP rupture/ debonding and concrete crushing.
εcu εfe + εf 0 εcuh = ⇒ xbb = xbb h − xbb εcu + εfe + εf 0
(6.11)
The balanced plate ratio for simultaneous FRP failure and concrete crushing is calculated from: ρf,bb =
0.405fcuxbb/d + 0.95(ρ′ε′s Es − ρfy) εfeEf
where ε′s = εcu −
(6.12)
d′
(εcu + εfe + εf 0) is assumed to be less than εy. If the comh pression steel has yielded, the term ρ′ε′s Es in Eq. (6.12) should be replaced by ρεy. 6.4.3.4 Checking for yielding of compression steel at ultimate state To calculate the moment at ultimate failure, it is necessary to know if the compression steel has yielded or not. As the plate ratio increases, the neutral axis moves downward and the strain in the compression steel increases. Therefore, compressive yielding of the steel will occur as long as the plate ratio goes beyond a certain critical value (ρf,cy). ρf,cy depends on the specific failure mode after steel yielding, and is derived below separately for (1) FRP rupture or debonding and (2) concrete crushing. (A) FRP RUPTURE OR DEBONDING
As illustrated in Figure 6.16, the depth of the neutral axis at this condition is given by:
Strengthening concrete structures with FRP
308
Figure 6.16 Determination of ρf,cy under two different situations.
εy εfe + εf0 εyh + (εfe + εf0)d′ = ⇒ xcy = xcy − d′ h − xcy εy + εfe + εf0
(6.13)
The critical plate ratio, beyond which compression steel yields at or before beam failure is obtained from equilibrium as: ρf,cy =
0.405 fcuxcy / d + 0.95(ρ′ − ρ)fy εfeEf
(6.14)
(B) CONCRETE CRUSHING
As illustrated in Figure 6.16, the depth of the neutral axis at this condition is given by: εy xcy − d′
=
εcu εcud′ ⇒ xcy = xcy εy + εcu
(6.15)
The critical plate ratio can then be derived from: ρf,cy =
0.405 fcuxcy/d + 0.95(ρ′ − ρ)fy εfEf
where εf = εcu
h − xcy − εf 0 ≤ εfe xcy
(6.16)
Strengthening concrete structures with FRP
309
6.4.3.5 Calculation of the ultimate moment The maximum moment that can be carried by a strengthened beam section depends on the mode of failure. In the following, equations for calculating the ultimate moment are provided separately for FRP rupture/debonding and concrete crushing failure. (A) FRP RUPTURE OR DEBONDING
(i) Compression steel yields at ultimate load (ρf ≥ ρf,cy) Mn = 0.95As fy (d − 0.45x) + Af εfe Ef (h − 0.45x) + 0.95A′s fy (0.45x − d′)
(6.17)
where: x=
0.95As fy + Af εfe Ef − 0.95 A′s fy 0.405 fcub
(ii) Compression steel does not yield at ultimate load (ρf < ρf,cy) Mn = 0.95Asfy(d − 0.45x) + Af εfe Ef (h − 0.45x) + 0.95A′s
x − d′
冢 h − x 冣 (ε
fe
+ εf 0) Es(0.45x − d′)
(6.18)
In this case, the depth of the neutral axis x is calculated using the following equation: ¯¯ x + C ¯¯ = 0 A¯x2 + B
(6.19)
where: A¯ = 0.405 fcub ¯¯ = −0.405 fcubh − 0.95(εfe + εf 0)A′s Es − 0.95As fy − Af εfe Ef B ¯¯ = 0.95(εfe + εf 0)Es A′s d′ + (0.95As fy + Af εfeEf )h C (B) CONCRETE CRUSHING
(i) Compression steel yields at ultimate load (ρf ≥ ρf,cy) Mn = 0.95Asfy(d − 0.45x) + Af + 0.95 A′s fy (0.45x − d′)
冢
h−x x
冣
εcu − εf 0 Ef(h − 0.45x) (6.20)
310
Strengthening concrete structures with FRP
where x is obtained from Eq. (6.19) using the following parameters: A¯ = 0.405fcub ¯¯ = 0.95(A′s − As)fy + (εcu + εf 0)EfAf B ¯¯ = −εcuhAfEf C (ii) Compression steel does not yield at ultimate load (ρf < ρf,cy) Mn = 0.95As fy(d − 0.45x) + Af
冢
冢
h−x x
冣
εcu − εf 0 Ef (h − 0.45x)
x − d′ Es(0.45x − d′) x
+ 0.95A′s εcu
冣
(6.21)
where x is calculated from Eq. (6.19) with the following parameters: A¯ = 0.405fcub ¯¯ = 0.95Esεcu A′s − 0.95As fy + (εcu + εf 0)Ef Af B ¯¯ = −εcuhAf Ef − 0.95εcud′A′s Es C Example calculation A reinforced concrete beam 400 mm in depth and 270 mm wide is strengthened by the wet lay-up of FRP plate, 0.334 mm in thickness and 250 mm in width, terminated on each side at 100 mm from the supports. The member is loaded under four-point bending with shear span of 1.3 m. The tension steel area is 900 mm2 at an effective depth of 340 mm, while the compression steel is 142 mm2 in area at a depth of 40 mm from the compressive surface. The concrete cylinder strength is 29.3 MPa. The steel Young’s modulus and yield strength are 200 GPa and 484 MPa respectively. The FRP Young’s modulus is 230 GPa and its rupture strain is 0.0148. Neglecting the initial strain when FRP is bonded, the ultimate moment of the strengthened member can be obtained as follows: (1) Compute the effective laminate strain: tpEp / d = 225.9N / mm2, bp / b = 0.926, d = 340mm, h/d = 1.176 S / d = 3.82, a / S = 0.077, fc′ = 29.3 N / mm2 (εfu)chart = 9940 × 10−6 Cbp / b = −0.274 · (0.926/0.8 − 1) + 1= 0.957 Cd = 0.0051 · (340/200 − 1)2 − 0.1453 · (340/200 − 1) + 1 = 0.901
Strengthening concrete structures with FRP
311
Ch / d = 0.987 · (1.176/1.2 − 1) + 1 = 0.981 CS / d = 0.1967 · (3.82/4 − 1) + 1 = 0.991 Ca / S = 0.0008 · (0.077/0.1 − 1)2 − 0.0627 · (0.077/0.1 − 1) + 1 = 1.015 Cfc′ = 0.2862 · (29.3/40 − 1) + 1 = 0.923 εpu = (9940 × 10−6 · 0.957 · 0.901 · 0.981 · 0.991 · 1.015 · 0.923)/1.6 = 4878 × 10−6 < 0.8 · 14800 × 10−6 = 8457 × 10−6 εfe = 4878 × 10−6/1.4 = 3484 × 10−6 (2) Compute ρf,b xb = =
εcu ·d εcu + εy 3500 3500 + 2420
· 340
= 201 mm In the calculations, take fcu = fc′ /0.8 = 36.6 MPa 201 − 40 0.0035 = 0.0028 > εy Strain in compressive steel (ε′s) = 201 = 0.00242 Stress in compressive steel = fy 0.405 · fcu · ρf,b =
xb d
+ 0.95 · (ρ′fy − ρfy ) Ef εf
εf = εcu ·
h − xb 400 − 201 = 0.0035 · = 0.003465 < εfe = 0.003484 xb 201
ρ = As / bd = 900/(270 · 340) = 0.0098 ρ′ = As′ / bd = 142/(270 · 340) = 0.0015 0.405 · 36.6 · ρf,b = ρf =
201 + 0.95 · (0.0015 − 0.0098) · 484 340 = 0.0062 230000 · 0.003465
Af 0.334 · 250 = = 0.00091 bd 270 · 340
ρf < ρf,b
to ensure the steel will yield
312
Strengthening concrete structures with FRP
(3) Compute ρf,bb εcu 3500 ·h= · 400 = 200 mm εcu + εfe + εfo 3500 + 3484
xbb =
ε′s = εcu −
d′ h
(εcu + εfe + εfo)
= 3500 × 10−6 −
40 · (3500 + 3484) × 10−6 400
= 2802 × 10−6 > εy
ε′s = εy
xbb + 0.95 · (ρ′ε′s Es − ρfy ) d Ef · εfe
0.405 · fcu ρf,bb =
0.405 · 36.6 · =
200
+ 0.95 · (0.0015 − 0.0098) · 484 340 230000 · 0.003484
= 0.00612 ρf < ρf,bb Therefore, debonding is the predominated failure mode. (4) Compute ρf,cy xcy = =
εy h + (εfe + εfo) · d′ εy + εfe + εfo 2420 · 400 + (3484 + 0) · 40 2420 + 3484 0.405 · fcu ·
ρf,cy =
xcy d
= 187.6
+ 0.95 · (ρ′ − ρ) · fy Efεfe
0.405 · 36.6 · = = 0.0054 ρf < ρf,cy
187.6 + 0.95 · (0.0015 − 0.0098) · 484 340 230000 · 0.003484
Strengthening concrete structures with FRP
313
Using Eq. (6.19) to compute: A¯ = 0.405 · fcu · b = 0.405 · 36.6 · 270 = 4002.2 ¯¯ = −0.405 · fcu · bh − 0.95 · εfe · As′Es − 0.95 · Asfy − Af εfe Ef B = −0.405 · 36.6 · 270 · 400 − 0.95 · 0.003484 · 200000 · 142 − 0.95 · 900 · 484 − 0.003484 · 230000 · 83.5 = −2.18 × 106 ¯¯ = 0.95 · (εfe + εfo) · EsA′s d′ + (0.95 · As fy + Af εfe Ef ) · h C = 0.95 · 0.003484 · 200000 · 142 · 40 + (0.95 · 900 · 484 + 83.5 · 0.003484 · 230000) · 400 = 1.96 × 108 ¯¯ − −B x=
冪B¯¯ − 4A¯C¯¯ 2
= 114 mm 2A¯ Mn = 0.95 · As fy (d − 0.45x) + Afεfe Ef(h − 0.45x) x − d′ + 0.95 · A′s (εfe + εf 0)Es(0.45x − d′) h−x
冢
冣
= 0.95 · 900 · 484 · (340 − 0.45 · 103.8) + 83.5 · 0.003484 · 230000 · (400 − 0.45 · 103.8) + 0.95 · 142 ·
冢400 − 103.8冣 · 0.003484 · 200000 · (0.45 · 103.8 − 40) 103.8 − 40
= 1.43 × 108 (N.mm) or 143 kNm 6.4.3.6 Serviceability considerations To avoid excessive cracking and deformation at the serviceability state, yielding of steel should be avoided. Following the suggestion of ACI(2002), the steel stress under service load should not exceed 80 percent of the yield strength. That is: fs,s ≤ 0.80 fy
(6.22)
To find the steel stress under service load, linear-elastic behavior is assumed for the compressive concrete. Analysis of the cracked section gives the following:
Strengthening concrete structures with FRP
314
fs,s =
[Ms + εf 0Af Ef (h − xe /3)](d − xe)Es AsEs(d − xe /3)(d − xe) + Af Ef(h − xe/3)(h − xe) + A′s Es(xe − d′)(xe /3 − d′) (6.23)
where Ms is the maximum moment associated with the service load, and xe is the depth of neutral axis obtained from:
1 2
bx2e +
冢E − 1冣A′ (x − d′) = E A (d − x ) + E A [h − 冢1 + ε 冣x ] Es
Es
s
e
Ef
s
c
c
e
εf0
f
c
e
(6.24)
c
6.4.3.7 Creep-rupture limits Under sustained loading, FRP materials may fail at a loading significantly below its short-term strength. This phenomenon, known as creep rupture, is most severe for GFRP and least for CFRP. Adverse environmental conditions such as high temperature, high alkalinity, wetting/drying and freezing/ thawing cycles may aggravate the reduction in strength. Following ACI (2002), the sustained stress levels in GFRP, AFRP and CFRP are limited to 20 percent, 30 percent and 55 percent of their respective short-term strength. Under continuous cyclic loading, the maximum stress during the cycle is also limited to the same value. As the sustained loading or the maximum load during continuous cyclic loading is unlikely to be very high, the FRP stress can be obtained in the same way as the steel stress in the above serviceability check. Indeed, once the steel stress is found from Eq. (6.23), the FRP stress can be obtained from: ff,s = fs,s
Ef h − xe
冢E 冣 d − x − ε
Ef
f0
s
(6.25)
e
6.5 Shearing strengthening of beams 6.5.1 Failure modes When the transverse steel reinforcements in an existing concrete beam are not able to provide sufficient shear capacity, shear strengthening can be performed with bonded FRP on the sides of the beam. When loading is applied, shear failure of the strengthened member starts with the formation of an inclined crack in the shear span. With steel stirrups and FRP bridging the crack and transferring stress back to the concrete, additional inclined cracks will form. The cracking pattern is best observed from a member strengthened with FRP strips, where cracks are revealed on the concrete surface between the strips. The opening of a crack will increase the FRP stress and initiate FRP debonding on its two sides. For concrete beams
Strengthening concrete structures with FRP
315
strengthened through side-bonding or U-jacketing of FRP, three different failure modes are possible. If the FRP strength is reached before debonding progresses to a free edge of the bonded FRP, FRP rupture is the failure mode. Conversely, debonding failure occurs if the FRP debonds completely from the concrete substrate before the occurrence of rupture at any location. Here, complete debonding means that the FRP debonds to such
Figure 6.17 (a) Debonding failure for a beam strengthened by U-jacketing; (b) FRP rupture for a beam strengthened by full wrapping
316
Strengthening concrete structures with FRP
an extent that it no longer carries any loading. For example, if a major shear crack intersects with a side-bonded FRP strip at a location 100 mm from its bottom edge, complete debonding is considered to occur if the lower 100 mm of the FRP has debonded from the concrete substrate. The third failure mode is a combination of the first two, with rupture of the FRP over part of the bonded area and complete debonding over other parts. For beams strengthened by full wrapping, final failure always occurs by FRP rupture. Even if the FRP on both sides of the beam has fully debonded from the concrete substrate, the anchorage provided at the top and bottom of the wrap allows the FRP stress to increase continuously until rupture occurs. Debonding failure and rupture failure are illustrated in Figure 6.17(a) and Figure 6.17(b) respectively. For both failure modes, failure is found to initiate at a location around the middle part of the shear span, rather than near the support or the loading point (for a specimen under four-point loading). Two reasons for this observation can be proposed. First, due to the formation of multiple inclined cracks along the shear span (Figure 6.18), FRP strips near the middle of the span are intersected by several cracks along their lengths, and hence more severely stretched than strips near the support or the loading point. Second, the crack opening, which governs FRP rupture or debonding, is not uniform along a crack. As the tips of inclined cracks are near the loading point, and opening near the crack tip is small, the FRP around the loading point is not heavily stretched. Near the support, the crack opening is controlled by the longitudinal reinforcements. Crack opening is therefore expected to be highest at the middle part of an inclined crack, which is also the location around the middle of the shear span. When rupture or complete debonding occurs, the FRP stress essentially drops to zero, so the ruptured or fully debonded parts no longer contribute to the load-carrying capacity. The stress in the remaining FRP is increased
Figure 6.18 Effect of multiple cracks on the straining and debonding of FRP strip around the middle of the shear span.
Strengthening concrete structures with FRP
317
and they may fail immediately or after additional loading is applied. An important point to note is that the FRP strips at different locations do not reach their ultimate stress (for either full debonding or rupture to occur) at the same time. In practical design, an effective failure strain (εfe), which represents the average FRP contribution at ultimate loading, should be employed. The determination of εfe will be discussed in Section 6.5.3. 6.5.2 Factors influencing the shear contribution of FRP reinforcement 6.5.2.1 Effect of FRP stiffness As discussed in the former section on flexural strengthening, FRP debonding is affected by the stiffness of the bonded sheet or strip. With increasing stiffness, it is easier for debonding to occur. Triantafillou (1998) suggested to represent the FRP stiffness by ρf Ef (the product of area fraction and Young’s modulus of the FRP), and argued that the effective FRP failure strain (εfe) should be dependent on this parameter. For failure due to complete debonding, it is well known that the maximum FRP stress (or strain) at failure decreases with stiffness of the FRP. For fiber rupture, failure in most cases will occur after significant (though incomplete) debonding along the interface. εfe, which is an averaged strain value, should be dependent on the extent of debonding at various parts of the FRP, which affects the stress distribution along the major shear crack. In Triantafillou (1998), εfe was plotted against ρfEf for a collection of experimental results, and data for both FRP rupture and complete debonding were found to fall along a single curve. 6.5.2.2 Effect of concrete strength FRP debonding is governed by the surface properties of the concrete, such as its surface tensile strength and interfacial fracture energy. These parameters are often not measured, but they can be related to the concrete compressive strength. In Khalifa et al. (1998) and Triantafillou and Antonopoulos (2000) εfe is taken to be proportional to f 2/3 c . The physical basis of this assumption lies in the fact that debonding is affected more by the tensile strength of concrete, which is related to the compressive strength through the 2/3 power relationship. In the ACI and FIB design equations for Ujacketing and side bonding, the parameter f 2/3 c appears. Obviously, a higher value of f 2/3 c leads to better interfacial bonding and more effective strengthening in such cases. In Triantafillou and Antonopoulos (2000), εfe was found to be dependent on (Ef ρf /fc2/3) for all strengthening configurations. Empirical equations to calculate the effective failure strain from (Ef ρf /fc2/3) were therefore proposed. These equations were later incorporated into the FIB design recommendations.
318
Strengthening concrete structures with FRP
6.5.2.3 Effect of strengthening configuration The strengthening configuration affects the failure mode. For both sidebonding or U-jacketing, complete debonding of FRP may occur, but the failure load is normally lower for side-bonded FRP, as there is no anchorage at either end of the FRP. For full wrapping, FRP rupture is the failure mode and the effective failure strain is higher than that for complete debonding. Therefore, in terms of strengthening effectiveness, full wrapping is the highest, followed by U-jacketing and then side-bonding. In practice, when Ujacketing and side-bonding are performed, it is desirable to apply anchorage to the free edge(s) of the FRP. The effectiveness of the anchorage should be verified by experiments. 6.5.2.4 Effect of steel shear reinforcement ratio For strengthened beams failed by complete debonding of side-bonded FRP strips, Pellegrino and Modena (2002) showed that the steel shear reinforcement has a significant effect on the effectiveness of shear strengthening. According to experimental findings, the effectiveness of shear strengthening decreases with an increase in the stiffness ratio between steel shear reinforcements and bonded FRP, given by ρs,f = EsAsv / Ef Af. Physically, the effect is attributed to the change in cracking pattern with increasing amount of steel stirrups in the beam. When there is no shear reinforcement in the beam, only a single major crack will form along the shear span. With more steel reinforcements to transfer stress back into the concrete, more cracks will form. When the FRP is intersected by a large number of opening cracks, complete debonding becomes easier to occur. 6.5.2.5 Effect of member size In an investigation by Leung et al. (2007), geometrically similar concrete beams of three different sizes (with depth of 180 mm, 360 mm and 720 mm) are strengthened with equal area fractions of FRP in both U-jacketing and full wrapping configurations to study the effect of size on the strengthening effectiveness. For full wrapping, the ratio of failure load between strengthened beam and control beam is similar for all specimen sizes, indicating minimum size effect. For U-jacketing, however, the strengthening effectiveness decreases significantly with member size. Further investigations are required to clarify and quantify the effect of size on shear strengthening. 6.5.2.6 Effect of shear span-to-depth ratio As shear failure of concrete beams occurs in different ways for different span-to-depth ratios, the effectiveness of shear strengthening is also expected to depend on the span-to-depth ratio. In the literature, shear
Strengthening concrete structures with FRP
319
strengthening with FRP bonding is found to be effective for concrete beams with various shear span-to-depth ratios ranging from 1 to 3, but a comprehensive study to quantify the effect of this parameter has yet to be performed. The results from such an investigation will be useful in refining existing design equations in the future. 6.5.3 Design of shear-strengthened concrete beams To find the shear capacity of a concrete beam strengthened with bonded FRP on the sides, the FRP can be treated in a similar way to steel shear reinforcements, and its contribution (Vf) is added to those from (1) the concrete and longitudinal steel reinforcements (Vc) and (2) the transverse steel reinforcements or bent-up bars (Vs). The total shear capacity (V) is then given by: V = Vc + Vs + Vf
(6.26)
Vc, which makes up of the shear resistance of concrete in compression, aggregate interlock along the shear crack as well as dowel action of the longitudinal steel reinforcements, is calculated with the same equations for conventional reinforced concrete design. In the calculation of Vs, all transverse steel reinforcements bridging the shear crack are assumed to have yielded. To find Vf, the effective FRP strain at failure (εfe ) needs to be obtained first. Equations for calculating εfe are given in design guidelines from ACI, FIB and JSCE. However, only the ACI equations account for the difference between full wrapping, U-jacketing and side bonding. In the FIB guideline, the same equation is proposed for U-jacketing and side bonding. In the JSCE guideline, only a single equation is proposed for all cases. In the following, the ACI equations are adopted. Treating the FRP sheet or stirrup in a similar way to steel stirrups, Vf is calculated from: Vf = Af Ef
冢
0.70εfe (sin α + cos α)df γf sf
冣
(6.27)
The various terms in Eq. (6.27) are illustrated in Figure 6.19. df is the effective length of the FRP, which is equal to the FRP length minus the length from the bottom of the beam to the centroid of the tensile steel reinforcements (see Figure 6.19(a)). In Figure 6.19(b), α is the principal fiber orientation for a FRP sheet or the inclination of a FRP strip (with fibers running along the strip). Also, θ is the inclination of shear crack which is taken to be 45 degrees. If FRP strips are used, the center to center spacing is given by sf (Figure 6.19(c)), and Af is equal to 2tf wf , where tf and wf are respectively the strip thickness and width. For a FRP sheet, (Af /sf ) in Eq. (6.27) is replaced by 2tf .
Strengthening concrete structures with FRP
320
Figure 6.19 Definition of terms in the design equations.
In Eq. (6.27), εfe is determined from empirical fitting of test data and the factor of 0.70 is incorporated to obtain a value with low probability of failure. Depending on application condition, FRP type and the failure mode, the corresponding material safety factor (γf) from Table 6.6 can be used. To find the mean effective failure strain (εfe ), equations from ACI (2002) are employed. For full wrapping or FRP that is properly anchored, failure occurs by FRP rupture and the effective failure strain is given by: εfe = min(0.75εfu,0.004)
(6.28)
In Eq. (6.28), εfu is the rupture strain of the FRP. The effective failure strain is limited to 0.004 to ensure that aggregate interlock is maintained when ultimate failure occurs. For U-jacketing or side bonding, εfe = min(κνεfu,0.75εfu,0.004)
(6.29)
where: k1k2Le 11900εfu
(6.30)
Le =
23300 (tf Ef )0.58
(6.31)
k1 =
冢27冣
κν = with
fc
2/3
⎧ df − Le ⎪ df k2 = ⎨ ⎪ df − 2Le ⎩ df
(6.32) for
U-Jacketing (6.33)
for
Side Bonding
Strengthening concrete structures with FRP
321
The design equations proposed above do not take into proper consideration the effects of span depth ratio, member size and steel reinforcement ratio on the FRP failure strain εfe. With additional experimental results available in the future, and better models developed to quantify the various effects, these equations will be refined. The calculation of shear capacity from Eqs (6.26) and (6.27) neglects the possibility of crushing of the compressive concrete struts between inclined cracks. To ensure the validity of this assumption, the ultimate shear capacity (V) should be limited to vubwd, with vu = 0.8 冪fcu but not exceeding 5 N/mm2. 6.5.4 Maximum FRP strip spacing For steel stirrups inside a concrete beam, the maximum FRP spacing is limited by the effective depth to the steel, in order for each inclined crack (assumed to run at 45 degrees to the longitudinal axis) to be intersected by at least one set of stirrups. For FRP strips, a smaller spacing is required to ensure that the crack is not just intersected by at least one FRP strip, but the FRP has to provide significant strengthening at the location of the intersection. For example, if a side-bonded FRP intersects with the crack at a location close to its free edge, complete debonding occurs easily and the effectiveness of the strip is very low. It is much more desirable to have the crack intersecting with the FRP at a location farther away from its edge. To ensure effective strengthening, Teng et al. (2002) proposed that there should be at least two FRP strips crossing a crack. The maximum strip spacing is then given by: sf ≤ sf, max =
df (sin α + cos α) df = 2 2
if α = 90°
(6.34)
where df is the effective depth of the FRP strip, as shown in Fig.6.19a.
6.6 Strengthening of concrete columns 6.6.1 Behavior of the FRP strengthened column and failure mode As discussed in Section 6.2.2, the strengthening of concrete columns relies on the confining effect of FRP on concrete when it cracks and expands laterally under increasing longitudinal compressive strain. To understand the behavior of FRP strengthened columns, it is informative to look at the compressive behavior of concrete under constant lateral confinement first. Here, only circular members with uniform confinement around its circumference are considered. Rectangular members will be discussed in Section 6.6.2.3. Figure 6.20 shows the stress–strain behavior of unconfined concrete, concrete under uniform confinement, and concrete under continuously increasing confinement. As lateral deformation of concrete increases
322
Strengthening concrete structures with FRP
Figure 6.20 The stress–strain behavior of unconfined concrete and concrete under constant and continuously increasing confinement.
with axial strain, FRP-confined concrete belongs to the last category. For unconfined concrete, the stress drops rapidly after the peak value, showing a brittle failure mode. When constant confinement is applied, there is an increase in both the concrete strength and the strain at which the strength is reached. The post-peak load-carrying capacity is greatly improved over that of unconfined concrete, but increasing strain is accompanied by a softening behavior (i.e. decrease in stress). For FRP-confined concrete, the lateral confinement increases continuously with applied axial strain. With increasing strain, the stress value is moving from one constant confinement curve to the next with higher confinement. As illustrated in Figure 6.20, this results in a hardening behavior of the confined column, which is commonly observed in experimental investigations on confined circular columns. In the hardening regime, significant lateral expansion of the member can be observed. Final failure occurs due to FRP rupture (Figure 6.21). Once failure occurs, there is a rapid drop in the load-carrying capacity. Ultimate failure is therefore brittle, but the large deformation before final failure provides sufficient warning and enables effective load redistribution among structural members. It should be pointed out that the confining effect of FRP is significant only at strain levels approaching or beyond the failure strain (i.e. the strain when the strength is reached) of the unconfined concrete. For concrete under uniaxial compression, splitting cracks tend to form in a direction parallel to the loading direction when the strength is approached. The corresponding increase in lateral strain activates the confining effect of the FRP. Before these cracks are formed, the lateral expansion is too small to induce high confining pressure from the FRP.
Strengthening concrete structures with FRP
323
Figure 6.21 Failure of an FRP-confined circular concrete specimen by FRP rupture.
6.6.2 Factors influencing the confining effect 6.6.2.1 Effect of FRP strength and thickness In this sub-section and the next, only circular columns are considered. Here, we further assume that the column is fully wrapped with FRP along its length (as in Figure 6.3(a)). When the column expands laterally, the relation between the confinement pressure and the stress in the FRP can be obtained from force equilibrium over half of the confining FRP, as illustrated in Figure 6.22. With D being the column diameter and tf the thickness of FRP, the confining pressure f is given by: f=
2σf tf ρfσf = D 2
(6.35)
where σf is the stress in the FRP and ρf is the area of FRP divided by the area of concrete. The maximum confinement fl is reached when the stress in the FRP reaches its strength (ffu). The maximum confinement is then given by: fl =
2ffu tf ρf ffu = D 2
(6.36)
324
Strengthening concrete structures with FRP
Figure 6.22 Relation between FRP stress and confining pressure.
It is obvious that the performance of the FRP-confined column depends on the maximum confining stress that can be provided. The strength of the FRP and its thickness (or area ratio) are hence important parameters governing the effectiveness of strengthening. 6.6.2.2 Effect of strengthening configuration Besides full wrapping, it is also possible to strengthen a column with discrete or continuous helical strips. In these cases, some regions of the column surface (specifically, the part between the strips) are not confined. However, even in these regions, confining stresses can still be present at a distance from the surface, due to the spreading of surface compression from the strengthened region towards the interior of the member (Figure 6.23). In this case, failure is governed by the section with the least confinement, which is at the middle of the unconfined region. The effective maximum confining stress in this section is reduced from 2ffutf /D by two factors, the first one accounting for the stress reduction when the confining stress spreads from the confined regions to cover the whole length of the column, and the second one accounting for the presence of unconfined concrete in the critical section (near the free surface). The first correction factor is given by bf/(bf +s′), with bf being the width of the FRP strip and s′ the clear spacing between adjacent strips. The second correction factor (ke) is given by (FIB 2001):
冢
ke = 1 −
s′ 2D
冣
2
(6.37)
When helical strips with pitch p are used, an additional correction factor (kh) is required to account for the increasing radius of curvature of a helix compared to that of the circular section. Following FIB (2001), the factor is:
Strengthening concrete structures with FRP
325
Figure 6.23 Confinement of concrete column by discrete FRP strips.
冢
kh = 1 + (
p 2 ) πD
冣
−1
(6.38)
6.6.2.3 Effect of member section geometry While FRP confinement is most effective for circular columns, rectangular columns are commonly employed in structures. When the rectangular column starts to crack and expand laterally, significant pressure only exists at the corners of the member. To strengthen the member, the proper rounding of the corners is crucial. If the corners are not rounded, the high stress concentration will result in pre-mature FRP rupture. Moreover, even if the FRP does not rupture, compressive force is only acting over a very small region of the surface at the vicinity of the corners. In this case, most of the concrete section remains unconfined, and the strengthening is minimal. With a larger radius of curvature at the corners, higher strengthening effectiveness can be achieved. The radius, however, is limited by the presence of reinforcements (especially steel hoops) within the concrete cover. In many cases, even though the strength improvement with FRP wrapping is very small, the post-peak stress softening occurs much slower than the unconfined member. The improved post-peak performance enables stress redistribution to other members and enhances the energy absorption capacity of the member.
326
Strengthening concrete structures with FRP
To find the failure load of rectangular columns wrapped by FRP, the effective confined area over the concrete section needs to be determined. This is beyond the scope of the present chapter and the interested reader can refer to Teng et al. (2002) for a thorough discussion. Also, since this is an ongoing research topic, major journals and conference proceedings in the field can be consulted for the most recent findings. 6.6.3 Design of FRP-strengthened column As shown in Figure 6.20, the confined circular column shows similar behavior to the unconfined member at the early stage of loading (up to the unconfined concrete strength) and then exhibits a hardening behavior with increasing strain. In design, the stress vs strain relation of the confined concrete can be approximated by a bilinear relation, as illustrated in Figure 6.24. Since the confinement effect is only significant after the concrete is stressed beyond its unconfined strength (fco′ ), the first branch of the bilinear relation can be taken as the line from zero to (ε′co, fco′ ), where ε′co is the strain corresponding to the unconfined strength. The second branch is then the line from (ε′co, fco′ ) to (ε′cc, fcc′ ), with fcc′ and ε′cc being the strength of confined concrete and the strain at ultimate failure. To obtain fcc′ and ε′cc, both FIB (2001) and ACI (2002) recommend using the empirical equation of Mandar et al. (1988), which was originally proposed for concrete under constant confinement, but was found to be applicable to FRP-strengthened columns as well. Using fl from Eq. (6.36)
冤
冪
fcc′ = fco′ 2.25 1 + 7.9
ε′cc =
1.71(5 fco′ − 4 fco′ ) Ec
fl fl − 2 − 1.25 fco′ fco′
冥
(6.39)
(6.40)
As the real stress–strain curve is convex, the use of a bilinear relation will always provide a conservative estimate. Using the bilinear relation, the behavior of a FRP-confined column under combined axial loading and bending can be analyzed with the same approach adopted in conventional reinforced concrete design. If a more accurate description of the stress vs strain relation is required, a procedure proposed by Spoelstra and Monti (1999), described in FIB (2001), can be followed. In design, partial safety factors have to be applied to the terms in Eq. (6.39). In the term (fl /fco′ ), the factor only needs to be applied to fl to account for the plausible reduction in FRP strength. Then, after fcc is obtained, it should be reduced by the factor of 1.5 as for unconfined concrete.
Strengthening concrete structures with FRP
327
Figure 6.24 A bilinear model to describe the complete stress–strain relation of FRPconfined circular concrete columns.
6.7 Summary First developed in the 1980s, FRP bonding is now a widely employed strengthening technique for concrete structures. In this chapter, after discussing the need of the technique, the FRP material and repair procedures are described. The applications of bonded FRP to the flexural and shear strengthening of beams as well as the compressive strengthening of columns are then covered. For each application, the failure behavior of the strengthened member and factors affecting the behavior are discussed. Equations for practical design are then given. The reader will notice that the section on flexural strengthening of beams is more thorough than those on the other two applications. This is because the current understanding of flexural strengthening with bonded FRP is beyond that for shear strengthening and compressive strengthening. The effects of various factors such as member size, span/depth ratio and steel reinforcement ratio on the effectiveness of shear strengthening, as well as the modeling of compressive behavior of rectangular members wrapped with FRP, are topics of ongoing research. The interested reader should consult recent issues of major journals and conference proceedings for the most up to-date information.
Bibliography
Abdelrazig, B.E.I., Sharp, J.H. and Jazairi, B.E. (1988) “The Chemical Composition of Mortars Made from Magnesia-Phosphate Cement”, Cement and Concrete Research, 18: 415–425. Achenbach, J.D., Komsky, I.N., Lee, Y.C. and Angel, Y.C. (1992) “Self-calibrating Ultrasonic Technique for Crack Depth Measurement”, Journal of Nondestructive Evaluation, 11: 103–108. ACI 201.2R-92 (1994a) “Guide to Durable Concrete”, in ACI Manual of Concrete Practice, Part I: Materials and General Properties of Concrete, Detroit, MI: ACI. ACI 209R-92 (1994b) “Prediction of Creep, Shrinkage, and Temperature Effects in Concrete Structures”, in ACI Manual of Concrete Practice, Part I: Materials and General Properties of Concrete, Detroit, MI: ACI. ACI 318–95 (1995) Building Code Requirements for Structural Concrete, Detroit, MI: ACI. ACI 201.1R-92 (1997) Guide for Making a Condition Survey of Concrete in Service, Detroit, MI: American Concrete Institute. ACI 305R-99 (1999) “Hot Weather Concreting”, in ACI Manual of Concrete Practice, Part 2, Detroit, MI: ACI. ACI 440.2R-02 (2002a) Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete Structures, Farmington Hills, MI: American Concrete Institute. ACI-440 (2002b) Report – Guide for the Design and Construction of Externally Bonded Frp Systems for Strengthening Concrete Structures. Detroit, MI: ACI. ACI 437R-03 (2003) Strength Evaluation of Existing Concrete Buildings, Detroit, MI: ACI. ACI 546R-04 (2004) Concrete Repair Guide, Detroit, MI: ACI. ACI (2006) Vision 2020: A Vision for the Concrete Repair, Protection, and Strengthening Industry, Farmington Hills, MI: ACI Strategic Development Council. ACI 216.1–07/TMS-0216–07 (2007a) Code Requirements for Determining Fire Resistance of Concrete and Masonry Construction Assemblies, Detroit, MI: ACI. ACI 224.1R-07 (2007b) Causes, Evaluation, and Repair of Cracks in Concrete Structures, Detroit, MI: ACI. ACI 349.1R-07 (2007c) Reinforced Concrete Design for Thermal Effects on Nuclear Power Plant Structures, Detroit, MI: ACI. ACI 364.1R-07 (2007d) Guide for Evaluation of Concrete Structures before Rehabilitation, Detroit, MI: ACI.
Bibliography 329 ACI Committee 201 (1984) Guide for Making a Condition Survey of Concrete in Service, ACI 201.IR-68, Detroit, MI: ACI. ACI Committee 228 (1998) Nondestructive Test Methods for Evaluation of Concrete in Structures, ACI 228.2R-98, Farmington Hills, MI: American Concrete Institute. ADWR (2006) Arizona Department of Water Resource, http://www.azwater.gov Ai, H. (2000) “Investigation of the Dimensional Stability in DSP Cement Paste”, PhD thesis, University of Illinois at Urbana-Champaign. Air Force (1994) Bridge Inspection, Maintenance, and Repair, Chapter 13: Concrete Bridge Maintenance, Repair, and Upgrade, Army TM 5–600, Air Force AFJPAM 32–1088. Allen, R.T.L., Edwards, S.C. and Shaw, J.D.N. (1993) The Repair of Concrete Structures, 2nd edn, Glasgow: Blackie. Alonso, C. and Andrade, C. (1994) “Life Time or Rebars in Carbonated Concrete”, in J.M. Costa and A.D. Mercer (eds), Progress in Understanding and Prevention of Corrosion, London: Institute of Materials. Alonso, C. Andrade, C. and Gonzalez, J. (1988) “Relation between Resistivity and Corrosion Rate of Reinforcements on Carbonated Mortar Made with Several Cement Types”, Cement and Concrete Research, 18: 687–698. Amparano, F.E. and Xi, Y. (1998) “Pumpability of Nonsand Concrete with AntiSegregative Additives”, ACI Materials Journal, 95(6): 695–703. Anonymous (1988) “Concrete Durability: A Multibillion-Dollar Opportunity”, Concrete International, 10(1): 33–35. ASCE Standard 7–95 (1995) Minimum Design Loads for Buildings and Other Structures, New York: American Society of Civil Engineers. ASCE (2000) Guideline for Structural Condition Assessment of Existing Buildings, SEI/ASCE 11–99, ASCE. ASTM C 876–91 (1999) “Standard Test Method for Half-Cell Potentials of Uncoated Reinforcing Steel in Concrete”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM C 1383 (2002) “Standard Test Method for Measuring the P-Wave Speed and Thickness of Concrete Plates Using the Impact-Echo Method”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM Test Designation C597–02 (2003a) “Standard Test Method for Pulse Velocity through Concrete”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM E 1316–02a (2003b) “Standard Terminology for Nondestructive Examinations, Section B: Acoustic Emission”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM (2004a) “Standard Specification for Epoxy-Coated Steel Reinforcing Bars, ASTM A775/A775M-04a, 01.04”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM (2004b) “Standard Test Method for Determining the Apparent Chloride Diffusion Coefficient of Cementitious Mixtures by Bulk Diffusion, C1556–04”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM C42 (2004c) “Standard Method of Obtaining and Testing Drilled Cores and Sawn Beams of Concrete”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM.
330
Bibliography
ASTM C1012–04 “Standard Test Method for Length Change of Hydraulic-Cement Mortars Exposed to a Sulfate Solution”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM C227–03 “Standard Test Method for Potential Alkali Reactivity of CementAggregate Combinations (Mortar-Bar Method)”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. ASTM C586–05 “Standard Test Method for Potential Alkali Reactivity of Carbonate Rocks as Concrete Aggregates (Rock-Cylinder Method)”, in Annual Book of ASTM Standards, West Conshohocken, PA: ASTM. Attoh-Okine, N. (1995) “Use of Artificial Neural Networks in Ground Penetrating Radar Applications in Pavement Evaluation and Assessment”, in G. Schickert, and H. Wiggerhanser (eds), International Symposium Non-Destructive Testing in Civil Engineering, Sept. 26–28, Berlin, pp. 93–100. Austin, S., Robins, P. and Pan, Y. (1999) “Shear Bond Testing of Concrete Repairs”, Cement and Concrete Research, 29: 1067–1076. Bartlett, F.M. and MacGregor, J.G. (1994) “Effect of Moisture Condition on Concrete Core Strengths”, ACI Materials Journal, 91(3): 227–236. Bazant, Z.P. (1979) “Physical Model for Steel Corrosion in Concrete Sea Structures”, Journal of the Structural Division, ASCE, 105(6): 1137–1166. Bazant, Z. P. (1992) “Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 6 – Simplified Code-Type Formulation”, Materials and Structures, 25(148): 219–223. Bazant, Z.P. and Kaplan, M.F. (1996) Concrete at High Temperatures, Material Properties and Mathematical Models, Harlow: Longman Group Limited. Beattie, A.G. (1983) “Acoustic Emission, Principal and Instrumentation”, Journal of Acoustic Emission, 2(1/2): 95–128. Beddoe, R. and Setzer, M.J. (1988) “A Low Temperature DSC Investigation of Hardened Cement Paste Subjected to Chloride Action”, Cement and Concrete Research, 18: 249–251. Beddoe, R. and Setzer, M.J. (1990) “Phase Transformations of Water in Hardened Cement Paste: A Low Temperature DCS Investigation”, Cement and Concrete Research, 20: 236–242. Bennett, E.W. and Loat, D.R. (1970) “Shrinkage and Creep of Concrete as Affected by the Fineness of Portland Cement”, Magazine of Concrete Research, 22(71): 69–78. Bentur, A., Diamond, S. and Berke, N.S. (1997) Steel Corrosion in Concrete: Fundamentals and Civil Engineering Practice, London: E & FN Spon. Berke, N.S., Pfeifer, D.W. and Weil, T.G. (1988) “Protection against ChlorideInduced Corrosion”, Concrete International Design & Construction, 10(12): 45–55. Bertolini, L., Elsener, B., Pedeferri, P. and Polder, R. (2004) Corrosion of Steel in Concrete: Prevention, Diagnosis and Repair, Berlin: Wiley-VCH Verlag GmbH & Co. KGaA. Bijen, J. (2003) Durability of Engineering Structures: Design, Repair and Maintenance, Cambridge: Woodhead Publishing Limited. Bilger, W., Murphy, W.E. and Reidinger, J. (1986) Inspection and Maintenance of Reinforced and Prestressed Concrete Structures, FIP Guide to Good Practice, London: Thomas Telford,
Bibliography 331 Bray, D.E. and McBride, D. (1992) Nondestructive Testing Techniques, New York, Wiley. Bray, D.E. and Stanley, R.K. (1997) Nondestructive Evaluation: A Tool in Design, Manufacturing and Service, Boca Raton, FL: CRC Press Brendenberg, H. (ed.) (1980) Proceedings of International Seminar on the Application of Stress-Wave Theory on Piles, Stockholm, June. Broomfield, J.P. (1997) Corrosion of Steel in Concrete: Understanding, Investigation and Repair, London: E & FN Spon. Broomfield, J.P., Rodriguez, J., Ortega, L.M. and Garcia, A.M. (1995) “Field Measurement of the Corrosion Rate of Steel in Concrete Using a Microprocessor Controlled Unit with a Monitored Guard Ring for Signal Confinement”, in Proceedings of Techniques to Assess the Corrosion Activity of Steel-Reinforced Concrete Structures, ASTM, STP 1276, Philadelphia. BS1881: Part 120 (1983) Method for Determination of the Compressive Strength of Concrete Cores, London: British Standards Institute. Bungey, J.H. and Millard, S.G. (1996) Testing of Concrete in Structures, 3rd edn, London: Blackie Academic & Professional. CAC Cement Association of Canada (2006), . Calleja, J. (1980) “Durability”, in Proceedings, 7th International Congress on the Chemistry of Cement, Paris, 1, pp. 2/1–2/48. Campbell-Allen, D. and Roper, H. (1991) Concrete Structures: Materials, Maintenance, and Repair, Harlow: Longman Scientific & Technical Carino, N.J., Sansalone, M. and Hsu, N.N. (1986) “A Point Source-Point Receiver Technique for Flaw Detection in Concrete”, Journal of American Concrete Institute, 83(2): 199. Champion, S. (1961) Failure and Repair of Concrete Structures, New York: John Wiley & Sons Inc. Chandler, I. (1991) Repair and Renovation of Modern Buildings, New York: McGraw-Hill Inc. Chandler, K.A. and Bayliss, D.A. (1985) Corrosion Protection of Steel Structures, Oxford: Elsevier Science Publishers. Chen, W.F. and Duan, L. (2000) Bridge Engineering Handbook, Cambridge: CRC Press. Chen, W.F. and Scawthorn, C. (2003) Earthquake Engineering Handbook, ICBO, SEA and CRC Press. Cherry, B.W. (1989) “Cathodic Protection of Underground Reinforced Concrete Structures”, paper presented at Institution of Corrosion Scientists and Technologists Second International Conference on Cathodic Protection Theory and Practice, June. Chew, M.Y.L. (1993) “Effect of Heat Exposure Duration on the Thermoluminescence of Concrete”, ACI Materials Journal, 90(4): 319–322. Clear, K.C. (1992) “Effectiveness of Epoxy-Coated Reinforcing Steel”, Concrete International, 14(5): 58–64. Clifton, J.R., Beeghley, H.F. and Mathey, R.G. (1975) “Nonmetallic Coatings for Concrete Reinforcing Bars”, Building Science Series, 65, US Department of Commerce, National Bureau of Standards, August. Colotti, V. and Spadea, G. (2001) “Shear Strength of RC Beams Strengthened with Bonded Steel or FRP Plates”, Journal of Structural Engineering, ASCE, 127(4): 367–373.
332
Bibliography
Concrete Society (1979) Specification for Sprayed Concrete, London: The Concrete Society. Concrete Society (1980) Code of Practice for Sprayed Concrete, London: The Concrete Society. Concrete Society (1981) Guidance Notes on the Method of Measurement for Sprayed Concrete, London: The Concrete Society. Concrete Society (1984) Repair of Concrete Damaged by Reinforcement Corrosion, London: The Concrete Society. Concrete Society (1987) Concrete Core Testing for Strength, London: The Concrete Society. CRA (2006) see Concrete Repair Association Home Page, http:// www.concreterepair.org.uk Crom, T.R. (1986) “Shotcrete Nozzleman Certification”, Concrete International, 8(2): 27–30. CSA (1989) “Report of Technical Committee on Repair of Concrete Buildings”, Draft 1. Davidson, N., Padaratz, I. and Forde, M. (1995) “Quantification of Bridge Scour Using Impulse Radar”, in G. Schickert and H. Wiggerhanser (eds), International Symposium on Non-Destructive Testing in Civil Engineering, Sept. 26–28, Berlin, pp. 61–68. Davis, H.E. (1940) “Autogenous Volume Change of Concrete”, Proceedings, ASTM, 40, pp. 1103–1110. Dawson, J.L., John, D.G., Jafar, M.I., Hladky, K. and Sherwood, L. (1990) “Electrochemical Methods for Inspection and Monitoring of Corrosion of Reinforcing Steel in Concrete”, in Proceedings of Corrosion of Reinforcement in Concrete, Oxford: Elsevier Applied Science, pp. 358–371. Dowrick, D. (2003) Earthquake Risk Reduction, New York: Wiley. Durand, B. and Gravel, C. (1995) “Evaluation of the Suitability of Lithium Hydroxide to Inhibit Expansion due to AAR both in New and Old Concrete Structures”, in Proceedings of the Second International Conference on AAR in Hydroelectric Plants and Dams, Tennessee, Oct., pp. 507–528. Emmons, P.H. (1994) Concrete Repair and Maintenance Illustrated, New York: R.S. Means Company. Emmons, P.H. (2006) “Concrete: Understanding When and Why Repairs are Necessary.” Buildings. http://www.buildings.com. ENV 1992-1-2 (1995a) Design of Concrete Structures – Part 1–2: General Rules – Structural Fire Design, Brussels: European Committee for Standardization. ENV 1993-1-2 (1995b) Design of Steel structures – Part 1–2: General Rules – Structural Fire Design, Brussels: European Committee for Standardization. Felicetti, R., Gambarova, P.G. and Meda, A. (2004) “Guidelines for the Structural Design of Concrete Buildings Exposed to Fire”, Workshop fib Task Group 4.3.2, Fire Design of Concrete Structures, Politecnico di Milano, Dec. 2–4. FHWA (1990) Hydraulic Engineering Circular (HEC) 18: Evaluating Scour at Bridges, Washington, DC: U.S. Department of Transportation, FHWA (1999) Material and Procedures for Rapid Repair of Partial-Depth Spalls in Concrete Pavements, Manual of Practice, FHWA Report No. FHWA-RD99–152, Washington, DC: U.S. Federal Highway Administration. FIB (2001) “Technical Report on Externally Bonded FRP Reinforcement for RC
Bibliography 333 Structures”, Bulletin No. 14, International Federation for Structural Concrete, Lausanne, Switzerland. FIP (1986) Inspection and Maintenance of Reinforced and Prestressed Concrete Structures, London: Thomas Telford. FIP (1991) FIP Guide to Good Practice of Repair and Strengthening of Concrete Structures, London: Thomas Telford. Folliard, K.J., Thomas, M.D.A. and Kurtis, K.E. (2003) Guidelines for the Use of Lithium to Mitigate or Prevent Alkali-Silica Reaction (ASR), FHWA Report No. FHWA-RD-03–047. Glass, G.K., Page, C.L. and Short, N.R. (1991) “Factors Affecting the Corrosion Rate of Steel in Carbonated Mortars”, Corrosion Science, 32: 1283–1294. Godfrey, J.R. (1984) “New Tools Help Find Flaws”, Civil Engineering, 54(9): 34–41. Gouda, V. K. (1970) “Corrosion and Corrosion Inhibition of Reinforcing Steel, I: Immersed in Alkaline Solutions”, British Corrosion Journal, 5(9): 198–203. Greene, G.W. (1954) “Test Hammer Provides New Method of Evaluating Hardened Concrete”, ACI J. Proc., 51(3): 249. Gutenberg, B. and Richter, C.F. (1956) “Magnitude and Energy of Earthquakes”, Ann. Geofis., 9: 1–15. Halicka, A. and Krol, M. (1999) “Evaluation and Testing of Bond Strength Between Ordinary and Expansive Concrete”, in R.K. Dhir and M.J. McLarthy (eds), Concrete Durability and Repair Technology, London: Thomas Telford, pp. 493–501 Hall, J.F. (1999) “Discussion on ‘the Role of Damping in Seismic Isolation’ ”, Earthquake Engineering and Structural Dynamics, 28: 1717–1720. Hall, J.F. and Ryan, K.L. (2000) “Isolated Buildings and the 1997 UBC Near-Source Factors”, Earthquake Spectra, 16: 393–411. Halpin, J.C. (1984) Primer on Composite Materials: Analysis, Technomic Co. Hanks, T.C. and Kanamori, H. (1979) “A Moment Magnitude Scale”, J. Geophys. Res., 84: 2348–2350. Hartt, W.H. and Rosenberg, A.M. (1989) “Influence of Ca(NO2)2 on Seawater Corrosion of Reinforcing Steel in Concrete”, American Concrete Institute, Detroit, SP 65–33, 609–622. Hausman, D. A. (1967) “Steel Corrosion in Concrete”, Materials Protection, 6(11): 19–22. Herakovich, C.T. (1998) Mechanics of Fibrous Composites, New York: John Wiley & Sons. Hisano, M., Nagashino, I. and Kawamura, S. (1990) “Earthquake Observation of a Base-Isolated Building with Slide Bearings”, in Proc. 8th JEES. Holl, C.H. and O’Connor, S.A. (1997) “Cleaning and Preparing Concrete Before Repair”, Concrete International, March, 60–63. Holland, T.C. (1983) Abrasion-Erosion Evaluation of Concrete Mixtures for Stilling Basin Repairs, Kinzua Dam, Pennsylvania, Miscellaneous Paper No. SL-83–16 US Army Engineers Waterways Experiment Station, Vicksburg. Hope, B.B., Neville, A.M. and Guruswami, A. (1967) “Influence of Admixtures on Creep of Concrete Containing Normal Weight Aggregate”, in RILEM International Symposium on Admixtures for Mortar and Concrete, Brussels, Sept., pp. 17–32. Hu, Y-X., Liu, S-C. and Dong, W. (1996) Earthquake Engineering, London: E & FN Spon.
334
Bibliography
Huang, X., Birman, V., Nanni, A. and Tunis, G. (2004) “Properties and Potential for Application of Steel Reinforced Polymer (SRP) and Steel Reinforced Grout (SRG) Composites”, Composites, Part B, 36, 1, 73–82. ICC (2002) 2003 International Building Code, Country Club Hills, IL: International Code Council ICC (2006) 2006 International Existing Building Code, Country Club Hills, IL: International Code Council. ICRI (1997) Guide for Selecting and Specifying Materials for Repair of Concrete Structures, International Concrete Repair Institute, Technical Guideline No. 03733. ICRI (2002) Guide for Surface Preparation for the Repair of Deteriorated Concrete Resulting from Reinforcing Steel Corrosion, International Concrete Repair Institute, Technical Guideline No. 03730. ICRI (2006) International Concrete Repair Institute website Jessop, E.L., Ward, M.A. and Neville, A.M. (1967) “Influence of Water Reducing and Set-Retarding Admixtures on Creep of Lightweight Aggregate Concrete”, in RILEM International Symposium on Admixtures for Mortar and Concrete, Brussels, Sept., pp. 35–46. Jones, R. (1962) Non-Destructive Testing of Concrete, Cambridge: Cambridge University Press. Jones, R.M. (1998) Mechanics of Composite Materials, 2nd edn, London: Taylor and Francis. JSCE (2001) Recommendations for Upgrading of Concrete Structures with Use of Continuous Fiber Sheets, Tokyo: Research Committee on Upgrading of Concrete Structures with Use of Continuous Fiber Sheets, Japan Society of Civil Engineers. Kahn, L.F. (2008) “ACI Code Requirements for Repair of Buildings”, Concrete International, April, 51–54. Kanamori, H. (1977) “The Energy Release in Great Earthquake”, J. Geophys. Res., 82: 2981–2987. Kay, T. (1992) Assessment and Renovation of Concrete Structures, New York: Longman Scientific & Technical. Keller, T. (2001) Use of Fibre Reinforced Polymers in Bridge Construction, State-ofthe-Art Report with Application and Research Recommendations, EPFL, Switzerland. Kelly, J.M. (1993) “The Role of Damping in Seismic Isolation”, Earthquake Engineering and Structural Dynamics, 28: 3–20. Kelly, J.M. (1997) Earthquake-Resistant Design with Rubber, New York: SpringerVerlag. Kesler, C.E. and Higuchi, Y. (1953) “Determination of Compressive Strength of Concrete by Using its Sonic Properties”, Proc. ASTM, 53: 1044. Kesler, C.E. and Higuchi, Y. (1954) “Problems in the Sonic Testing of Plain Concrete”, in Proc. Int. Symp. on Nondestructive Testing of Materials and Structures, 1, RILEM, Paris, p. 45. Keyser, J.H. (1980) Durability of Materials and Construction, ASTM STP691, 38. Khalifa, A., Gold, W.J., Nanni, A. and Aziz, A. (1998) “Contribution of Externally Bonded FRP to Shear Capacity of RC Flexural Members”, ASCE Journal of Composites for Construction, 2(4): 195–203. Klieger, P., Anderson, A.R., Bloem, D.L., Howard, E.L. and Schlintz, H. (1954)
Bibliography 335 “Discussion of ‘Test Hammer Provides New Method of Evaluating Hardened Concrete,’ ” by Gordon W. Greene, ACI J. Proc., 51(3): 256–251. Knab, L.I. (1988) Factors Related to the Performance of Concrete Repair Materials, Technical Report REMR-CS- 12,, Washington, DC: U.S. Army Corps of Engineers. Knab, L.I. and Spring, C. B. (1989) “Evaluation of Test Methods for Measuring the Bond Strength of Portland Cement Based Repair Material to Concrete”, Cement, Concrete and Aggregate, 11: 3–14. Kolek, J. (1958) “An Appreciation of the Schmidt Rebound Hammer”, Magazine of Concrete Research, 10(28): 27. Kolek, J. (1969) “Nondestructive Testing of Concrete by Hardness Methods”, in Proceedings of Symposium on NDT of Concrete and Timber, Institute of Civil Engineers, London, p.15. Kramer, S.L. (1996) Geotechnical Earthquake Engineering, Englewood Cliffs, NJ: Prentice Hall. Kunieda, M., Kurihara, N., Uchida, Y. et al. (2000) “Application of Tension Softening Diagrams to Evaluation of Bond Properties at Concrete Interfaces”, Engineering Fracture Mechanics, 65: 299–315. Landis, E.N. and Shah, S.P. (1993) “Recovery of Microcrack Parameters in Mortar Using Quantitative Acoustic Emission”, Journal of Nondestructive Evaluation, 12(4): 219–232. Lea, F.M. (1970) The Chemistry of Cement and Concrete, London: Arnold. Lee, J.S., Xi, Y. and Willam, K. (2008) “Properties of Concrete after High Temperature Heating and Cooling”, Journal of Materials, ACI, July–Aug. 334–341. Leung, C.K.Y. (2001) “Delamination Failure in Concrete Beams Retrofitted with a Bonded Plate”, ASCE Journal of Materials in Civil Engineering, 13(2): 106–113. Leung, C.K.Y. (2006) “FRP Debonding from a Concrete Substrate: Some Recent Findings against Conventional Belief”, Cement and Concrete Composites, 28(8): 742–748. Leung, C.K.Y., Chen, Z., Lee, S., Ng, M., Tang, J. and Xu, M. (2007) “Effect of Member Size on the Shear Failure of FRP Strengthened Reinforced Concrete Members”, ASCE Journal of Composites in Construction, 1(5): 487–496. L’Hermite (1960) Proceedings of the IVth International Symposium on the Chemistry of Cement, Washington, DC, 2, pp. 659–694. Li, Z. (1996) “Microcrack Characterization in Concrete under Uniaxial Tension”, Magazine of Concrete Research, 48(176): 219–228. Li, Z. and Li, W. (2002) “Contactless, Transformer-Based Measurement of the Resistivity of Materials”, United States Patent 6639401. Li, Z., Li, F.M., Li, X.S. and Yang W.L. (2000) “P-Wave Arrival Determination and AE Characterization of Concrete”, Journal of Engineering Mechanics, 126(2): 194–200. Li, Z., Li, F.M., Zdunek, A., Landis, E. and Shah, S.P. (1998) “Application of Acoustic Emission Technique to Detection of Reinforcing Steel Corrosion in Concrete”, ACI Materials Journal, 95(1): 68–76. Li, Z., Mu, B. and Peng, J. (1999) “The Combined Influence of Chemical and Mineral Admixtures Upon the Alkali-Silica Reaction”, Magazine of Concrete Research, 51(3): 163–169. Li, Z., Mu, B. and Peng, J. (2000) “Aklali-Silica Reaction of Concrete with
336
Bibliography
Admixtures- Experiment and Prediction”, Engineering Mechanics, ASCE, 126(3): 243–249. Li, Z., Peng, J. and Ma, B. (1999) “Investigation of Chloride Diffusion for High-Performance Concrete Containing Fly Ash, Microsilica, and Chemical admixtures”, ACI Materials Journal, 96(3): 391–396. Li, Z., Rossow, E.C. and Shah, S.P. (1989) “Sinusoidal Forced Vibration of Sliding Masonry System”, Journal of Structural Engineering, ASCE, 115(7): 1741–1755. Li, Z. and Shah, S.P. (1994) “Localization of Microcracking in Concrete Under Uniaxial Tensile”, ACI Materials Journal, July: 372–381. Li, Z., Wei, X. and Li, W. (2003) “Preliminary Interpretation of Portland Cement Hydration Process using Resistivity Measurements”, ACI Materials Journal, 100(3): 253–257. Li, Z., Yao, W., Lee, S., Lee, C.H. and Yang, Z.Y. (2000) “Application of Infrared Thermography Technique in Building Finish Evaluation”, Journal of Nondestructive Evaluation, 19(1): 11–19. Lin, J.M. and Sansalone, M. (1997) “A Procedure for Determining P-wave Speed in Concrete for Use in Impact-Echo Testing Using a Rayleigh Wave Speed Measurement Technique”, Innovations in Nondestructive Testing, SP-168, Detroit, MI: American Concrete Institute. Lin, J.M., Sansalone, M. and Streett, W.B. (1996) “A Procedure for Determining P-Wave Speed in Concrete for Use in Impact-Echo Testing Using a Direct P-wave Speed Measurement Technique”, ACI Material Journal, 93. Liu, J., Xie, H., Xiong, G. et al. (1998) “Using Ultra-Sonic Methods to Judge Quality the New-Old Concrete Interface”, in Proceedings of the 4th Academic Conference on National Building Examination and Reinforcement in China, Kunming, pp. 152–158 (in Chinese). Long, A.E. and Murray, A. (1984) The Pull-Off Partially Destructive Test for Concrete, ACI SP-82, pp. 327–350. Ludirdja, D., Berger, R.L. and Young, J.F. (1989) “Simple Method for Measuring Water Permeability of Concrete”, ACI Materials Journal, 86(5): 433–439. Mailvaganam, N.P. (1992) Repair and Protection of Concrete Structures, Boca Raton, FL: CRC Press. Mailvaganam N.P., Pye, G.B. and Arnott, M.R. (1998) “Surface Preparation of the Concrete Substrate”, Institute for Research in Construction, Maji, A.K., Ouyang, C. and Shah, S.P. (1990) “Fracture Mechanism of Quasi-Brittle Material Based on Acoustic Emission”, Journal of Material Research, 5(1): 206–217. Maji, A.K. and Shah, S.P. (1988) “Process Zone and Acoustic Emission Measurement in Concrete”, Experimental Mechanics, 28, 27–33. Maji, A.K. and Shah, S.P. (1989) “Application of Acoustic Emission and Laser Holography to Study Micro-Fracture in Concrete”, in Nondestructive Testing of Concrete, SP-112, Detroit: American Concrete Institute, pp. 83–109. Malek, A.M. and Patel, K. (2002) “Flexural Strengthening of Reinforced Concrete Flanged Beams with Composite Laminates”, ASCE Journal of Composites for Construction, 6(2): 97–103. Malhotra, H.L. (1956) “The Effect of Temperature on the Compressive Strength of Concrete”, Magazine of Concrete Research, 8: 85.
Bibliography 337 Malhotra, V.M. and Carino, N.J. (2004) Handbook of Nondestructive Testing of Concrete, 2nd edn, Boca Raton, FL: CRC Press. Mandar, J.B., Priestley, M.J.N. and Park, R. (1988) “Theoretical Stress-Strain Model for Confined Concrete”, ASCE Journal of Structural Engineering, 114(8): 1804–1826. Mangat, P.S. and Limbachiya, M.C. (1997) “Repair Material Properties for Effective Structural Application”, Cement and Concrete Research, 27: 601–617. Marosszeky, M., Yu, J.G. and Ng, C.M. (1991) “Prediction of Creep, Shrinkage, and Temperature Effects in Concrete Structures”, ACI Publication SP126–70 Durability of Concrete, Second International Conference, 2: 1331–1334. Mathey, R.G. and Knab, L.I. (1991) “Uniaxial Tensile Tested to Measure the Bond of In-Situ Concrete Overlays”, NISTIR 4648. Mays, G.C. (1985) “Structural Applications of Adhesives in Civil Engineering”, Materials Science and Technology, 1(11): 937–943. McKenna, J.K. and Erki, M.A. (1994) “Strengthening of Reinforced Concrete Flexural Members Using Externally Applied Steel Plates and Fibre Composite Sheets – A Survey”, Canadian Journal of Civil Engineering, 21(1): 16–24. McLeish, A. (1993) “Standard Tests for Repair Materials and Coatings for Concrete, Part I: Pull-Off Tests”, Technical Note 139, CIRIA. Mehta, P.K. (1980) “Durability of Concrete in Marine Environment – A Review”, ACI SP-65, pp. 1–20. Mehta, P.K. (1988a) “Sulfate Resistance of Blended Cements”, in W.G. Ryan (ed.) Papers, Concrete 88 Workshop Concrete Institute of Australia, pp. 337–351. Mehta, P.K. (1988b) “Durability of Concrete Exposed to Marine Environment – A Fresh Look”, in V.M. Malhotra (ed.), Concrete in Marine Environment, ACI SP-109, pp. 1–29. Mehta, P.K. and Gerwick, B.C. (1982) “Cracking-Corrosion Interaction in Concrete Exposed to Marine Environment”, Concete International, 4(10): 45–51. Mehta, P.K. and Monteiro, P.J.M. (2006) Concrete: Microstructure, Properties and Materials, 3rd edn, Maidenhead: McGraw-Hill. Meinheit, D.F. and Monon, J.F. (1984) “Parking Garage Repaired Using Thin Polymer Concrete Overlay Concrete”, Concrete International, 6(7): 7–13. Mindess, S., Young, J.F, and Darwin, D. (2003) Concrete, 2nd edn, New Jersey: Pearson Education. Monday, J.G.L. and Dhir, R.K. (1984) “Assessment of In-situ Concrete Quality by Core Testing”, Special Publication, SP 82–20, Detroit: American Concrete Institute, pp. 393–410. Nasser, K.W. and Al-Manaseer, A.A. (1987) “New Nondestructive Test”, ACI Concrete International, 41. National Research Council (1992) Techniques for Concrete Removal and Bar Cleaning on Bridge Rehabilitation Projects, Strategic Highway Research Program, SHRP-S-336, Washington, DC. National Research Council (1993a) Innovative Materials Development and Testing, Vol. 5: Partial Depth Spall Repair, Strategic Highway Research Program, SHRPH-356, Washington, DC. National Research Council (1993b) Concrete Bride Protection, Repair, and Rehabilitation Relative to Reinforcement Corrosion: A Methods Application Manual, Strategic Highway Research Program, SHRP-S-360, Washington DC, pp. 47–98.
338
Bibliography
NCPP (National Center for Pavement Preservation) (2004) Colorado DOT Distress Manual for HMA and PCC Pavements, Michigan State University, 2857 Jolly Road, Okemos, MI 48864. Netcomposites (2006) Neville, A.M. (ed.) (1975) “Fiber Reinforced Cement and Concrete”, in RILEM Symposium, London: Construction Press. Neville, A.M. (1996) Properties of Concrete, 4th edn., Harlow: Longman Group Limited. Neville, A.M., Dilger, W.H. and Brooks, J.J. (1983) Creep of Plain and Structural Concrete, London: Construction Press. Newman, A. (2001) Structural Renovation of Buildings: Methods, Details, and Design Examples, New York, McGraw-Hill Nielsen, C.V., Pearce, C.J., and Bicanic, N. (2004) “Improved phenomenological modeling of transient thermal strains for concrete at high temperature”, Computers and Concrete, March, 1(2): 189–209. Obert, L. and Duvall, W.I. (1941) “Discussions of Dynamic Methods of Testing Concrete with Suggestions for Standardization”, Proc. ASTM, 41, 1053. Olson. L.D. and Wright, C.C. (1990) “Seismic, Sonic, and Vibration Methods for Quality Assurance and Forensic Investigation of Geotechnical, Pavement, and Structural Systems”, in H.L.M. dos Reis (ed.), Prof. Conf. on Nondestructive Testing and Evaluation for Manufacturing and Construction, Hemisphere, p. 263. Ouyang C.S., Landis, E. and Shah, S.P. (1991) “Damage Assessment in Concrete Using Quantitative Acoustic Emission”, Journal of Engineering Mechanics, ASCE, 117(11): 2681–2698. Page, C.L. (1992) “Nature and Properties of Concrete in Relation to Reinforcement Corrosion”, in Corrosion of Steel in Concrete, Aachen, Feb. 17–19. Page, C.L. and Treadway, K.W.J. (1982) “Aspects of the Electrochemistry of Steel in Concrete”, Nature, 297: 109. Parsons Brinckerhoff (1993) Bridge Inspection and Rehabilitation: A Practical Guide, ed. L.G. Silano, New York: John Wiley & Sons, Inc. PCA, Portland Cement Association Home Page. 28. Pellegrino, C. and Modena, C. (2002) “Fiber Reinforced Polymer Shear Strengthening of Reinforced Concrete Beams with Transverse Steel Reinforcement”, ASCE Journal of Composites for Construction, 6(2): 104–111. Perkins, P.H. (1986) Repair, Protection and Waterproofing of Concrete Structures, London: Elsevier Applied Scientific Publishers Ltd. Phan, L.T. (2004) “Codes and Standards for Fire Safety Design of Concrete Structures in the U.S.”, in Workshop fib Task Group 4.3.2, Fire Design of Concrete Structures, Politecnico di Milano, Dec. 2–4. Placido, F. (1980) “Thermoluminescence Test for Fire Damaged Concrete”, Magazine of Concrete Research, 32(111): 112–116. Popovics, J.S., Song, W., Achenbach, J.D., Lee, J.H. and Andre, R.F. (1998) “One-sided Stress Wave Velocity Measurement in Concrete”, ASCE Journal of Engineering Mechanics, 124: 1346–1353. Popovics, S. (1987) “Chemical Resistance of Portland Cement Mortar and Concrete”, in W. L. Sheppard (ed.) Corrosion and Chemical Resistance Masonry Materials Handbook, New Jersey: Noyes Publications.
Bibliography 339 Powers, T.C., Copeland, L.E., Hayes, J.C. and Mann, H.M. (1954) “Permeability of Portland Cement Paste”, ACI Journal Proceedings, 51(3): 285–298. Pullar-Strecker, P. (1987) Corrosion Damage Concrete: Assessment and Repair, London: CIRIA Butterworths. Raoof, M. and Hassanen, M.A.H. (2000) “Peeling Failure of Reinforced Concrete Beams with Fibre-Reinforced Plastics or Steel Plates Glued to Their Soffits”, in Proceedings of the Institution of Civil Engineers: Structures and Buildings, 140: 291–305. Rhode Island State Building Code (1997) 7th edn, Providence, R.I. Richer, C.F. (1935) “An Instrumental Earthquake Scale”, Bulletin of the Seismological Society of America, 25: 1–32. Rizzo, E.M. and Sobelman, M. (1989) “Selection Criteria for Concrete Repair Materials”, Concrete International, 11(9): 46–49. Roberts, T.M. (1989) “Approximate Analysis of Shear and Normal Stress Concentrations in the Adhesive Layer of Plated RC Beams”, The Structural Engineer, 67(12): 229–233. Robins, P.J. and Austin, S.A. (1995) “A Unified Failure Envelope from the Evaluation of Concrete Repair Bond Tests”, Magazine of Concrete Research, 47: 57–68. Saccani, A. and Magnaghi, V. (1999) “Durability of Epoxy Resin-Based Materials for the Repair of Damaged Cementitious Composites”, Cement and Concrete Research, 29: 95–98. Sadegzadeh, M. and Kettle, R.J. (1988) “Abrasion Resistance of Surface Treated Concrete”, Cement, Concrete & Aggregates, 10(1): 20–28. Sansalone, M. (1997) “Impact-Echo: The Complete Story”, ACI Structural Journal, 6. Sansalone, M. and Carino, N.J. (1988) “Impact-Echo Method: Detecting Honeycombing, the Depth of Surface-Opening Cracks, and Ungrouted Ducts”, Concrete International, 10(4): 38–46. Sansalone, M. and Carino, N.J. (1989) “Detecting Delaminations in Concrete Slabs with and without Asphalt Concrete Overlays Using the Impact-Echo Method”, ACI Materials Journal, 86(2): 175–184. Sansalone, M. and Carino, N.J. (1990) “Finite Element Studies of the Impact-Echo Response of Layered Plates Containing Flaws”, in W. McGonnagle (ed.), International Advances in Nondestructive Testing, 15th edn, New York: Gordon & Breach Science Publishers, pp. 313–3336. Satake, J., Kamakura, M., Shirakawa, K., Mikami, N. and Swamy, R.N. (1983) “Long Term Resistance of Epoxy-Coated Reinforcing Bars”, in A. P. Crane (ed.), Corrosion of Reinforcement in Concrete Construction, New York: The Society of Chemical Industry/Ellis Horwood Ltd, pp. 357–377. Sebastian, W.M. (2001) “Significance of Midspan Debonding Failure in FRP-Plated Concrete Beams”, ASCE Journal of Structural Engineering, 127(7): 792–798. Seehra, S.S., Gupta, S. and Kumar, S. (1993) “Rapid Setting Magnesium Phosphate Cement for Quick Repair of Concrete Pavement”, Cement and Concrete Research, 23: 254–266. SEI/ASCE 11–99 (2000) ASCE Standard – Guideline for Structural Condition Assessment of Existing Buildings, Reston, Virginia: ASCE. Shiu, K.N. and Stanish, K. (2008) “Extending the Service Life of Parking Structures”, Concrete International, April, 43–49. Smith, L.L., Kessler, R.J. and Powers, R.G. (1993) “Corrosion of Epoxy Coated
340
Bibliography
Rebar in a Marine Environment”, Transportation Research Circular, 403, Transportation Research Board, National Research Council, pp. 36–45. Soares, J.B. and Tang, T. (1998) “Bimaterial Brazilian Specimen For Determining Interfacial Fracture Toughness”, Engineering Fracture Mechanics, 59(1): 57–71. Spoelstra, M.R. and Monti, G. (1999) “FRP-Confined Concrete Model”, ASCE Journal of Composites for Construction, 3(2): 143–150. Stehno, G. and Mall, G. (1977) “The Tear-Off Method: A New Way to Determine the Quality of Concrete in Structures on Site”, in Proceedings of RILEM International Symposium on Testing In-Situ of Concrete Structures, Budapest, pp. 335–347. Steinbach, J. and Vey, E. (1975) “Caisson Evaluation by Stress Wave Propagation Method”, Journal of Geotechnical Engineering, ASCE, 101(4): 361. Stokes, D.B. (2002) “The Role of Lithium to Mitigate ASR in Existing Concrete”, paper presented at the Workshop for Use of Lithium to Mitigate Alkali-Silica Reactivity, Transportation Research Board, 81st Annual Meeting, January 13, Washington, DC. Suaris, W. and van Mier, J.G.M. (1995) “Acoustic Emission Source Characterization in Concrete under Biaxial Loading”, Materials and Structures, 28(182): 444–449. Taylor, M.A. (1992) “Passive Acoustic Emission for Quantitative Evaluation of Freeze Thaw and Alkali Aggregate Reaction in Concretes”, in Nondestructive Testing of Concrete Elements and Structures, Proceedings of Sessions sponsored by the Engineering Mechanics Division of the American Society of Civil Engineers in Conjunction with the Structures Congress, San Antonio, TX, 13–15 April. Teng, J.G., Chen, J.F., Smith, S.T. and Lam, L. (2002) FRP Strengthened RC Structures. Chichester: John Wiley & Sons, Ltd. Tovey, A.K. (1986) “Assessment and Repair of Fire-Damaged Concrete Structuresan Update”, in T.Z. Harmathy (ed.) Evaluation and Repair of Fire Damage to Concrete, Detroit, MI: American Concrete Institute. Trabanelli, G. (1986) “Corrosion Inhibitors”, in F. Mansfeld (ed.), Corrosion Mechanism, New York: Marcel Dekker. Triantafillou, T.C. (1998) “Shear Strengthening of Reinforced Concrete Beams Using Epoxy-Bonded FRP Composites”, ACI Structural Journal, 95(2): 107–115. Triantafillou, T.C. and Antonopoulos, C.P. (2000) “Design of Concrete Flexural Members Strengthened in Shear with FRP”, ASCE Journal of Composites for Construction, 4(4): 198–205. Trout, J. (1997) Epoxy Injection in Construction, Hanley-Wood, Inc. Troxell, G.E, Davis, H.E, and Kelly, G.W. (1968) Composition and Properties of Concrete, 2nd edn, New York: McGraw-Hill. Utah DOT Research News (1998) “Clear Penetrating Sealers– the Concrete Protection of Tomorrow”, Number 98–2. Valenta, O. (1969) “Kinetics of Water Penetration into Concrete as an Important Factor of Its Deterioration and of Reinforced Corrosion”, RILEM International Symposium on the Durability of Concrete, Prague, Part I, pp. 177–193. Vassie, P.R. (1978) “Evaluation of Techniques for Investigating the Corrosion of Steel in Concrete”, Department of the Environment, Department of Transport, TRRL Report SR397, Crowthorne. Verbeck, G.J. (1975) “Mechanism of Corrosion in Concrete”, in Corrosion of Metals in Concrete, ACI SP-49.
Bibliography 341 Viktorov, I. (1967) Rayleigh and Lamb Waves, trans. W. P. Mason, New York: Plenum Press. Vuorinen, J. (1985) “Applications of Diffusion Theory to Permeability Tests on Concrete, Part I: Depth of Water Penetration into Concrete and Coefficient of Permeability”, Magazine of Concrete Research, 37(132): 145–152. Wall, J.S. and Shrive, N.G. (1998) “Factors Affecting Bond Between New and Old Concrete”, ACI Material Journal, 85: 117–125. Warner, J. (1984) “Selection of Repair Materials”, Concrete Construction – World of Concrete, 29(10): 865–871. Warner, J. (1997) “Methods of Repairing and Retrofitting (Strengthening) Existing Buildings”, paper presented at workshop on Earthquake-Resistant Reinforced Concrete Building Construction (ERCBC), University of California, Berkeley, Jul 11–15, 1977. Reprinted in ACI Seminar Course Manual, ACI Infrastructure Seminar – Rehabilitation of Concrete Structure (Undated), pp. 61–92. Weil, S. (1995) “Non-Destructive Testing of Bridge, Highway and Airport Pavements”, in G. Schickert, and H. Wiggerhanser (eds), International Symposium Non-Destructive Testing in Civil Engineering, Sept. 26–28, Berlin, pp. 467–474. Wood, H.O. and Neumann, F. (1931) “Modified Mercalli Intensity Scale of 1931”, Bulletin of the Seismological Society of America, 21: 277–283. Wyss, M. and Brune, J. (1968) “Seismic Moment, Stress and Source Dimensions”, Journal of Geophys. Res., 73: 4681–4694. Xanthakos, P.P. (1996) Bridge Strengthening and Rehabilitation, Upper Saddle River, NJ: Prentice Hall. Xi, Y. and Li, Y. (2004) “Materials, Testing Methods, and Construction Practices for Fast Concrete Deck Repair”, Internal Report CU/SESM/XI-2004/001 University of Colorado at Boulder, and Report to Colorado DOT. Xi, Y., Shing, B., Abu-Hejleh, N., Asiz, A., Suwito, Xie, Z.H., Ababneh, A. (2003) Assessment of the Cracking Problem in Newly Constructed Decks in Colorado, Colorado Department of Transportation, Report No. CDOT-DTD-R-2003–3. Xiong, G., Liu, J. and Li, G. (2002) “A Way of Improving Interfacial Transition Zone Between Concrete Substrate and Repair Materials”, Cement and Concrete Research, 32: 1–5. Yang, Q., Zhang, S. and Wu, X. (2000) “Deicer-Scaling Resistance of Phosphate Cement-Based Binder for Rapid Repair of Concrete”, Cement and Concrete Research, 30: 1807–1813. Yang, Q., Zhang, S. and Wu, X. (2002) “Deicer-Scaling Resistance of Phosphate Cement-Based Binder for Rapid Repair of Concrete”, Cement and Concrete Research, 32: 165–168. Yang, Q., Zhu, B., Zhang, S. and Wu, X. (2000) “Properties and Application of Magnesia-Phosphate Cement Mortar for Rapid Repair of Concrete”, Cement and Concrete Research, 30: 1807–1813. Yue, L.L. and Taerwe, L. (1992) “Creep Recovery of Plain Concrete and Its Mathematical Modelling”, Magazine of Concrete Research, 44(161): 281–290. Zhang, J., Qin L. and Li, Z. (2008) “Hydration Monitoring of Cement-Based Materials with Resistivity and Ultrasonic Methods”, accepted for publication, Materials and Structures. Zoldners, N.G. (1957) “Calibration and Use of Impact Test Hammer”, ACI J. Proc., 54(2): 161.
Index
Note: f indicates a figure and t a table abrasion 35–6 Achenbach, J.D. et al. 128 ACI (American Concrete Institute) 12, 19, 26, 42, 42t, 53, 53t, 82–3, 87, 147, 148, 149, 150–1, 196–7 acid attack 55–6 acid etching 184 acoustic emission (AE) 98–102, 99f, 144–6, 145f adhesive failure 171 ADWR 201 Alkali-Aggregate reaction (AAR) 47–51 Allen, R.T.L. et al. 14, 53, 57, 79, 80, 156, 184 Amparano, F.E. 190 Antonopoulos, C.P. 317 ASCE (American Society of Civil Engineers) 2, 78, 84, 85–6 ASR (alkali-silica reaction) 47–51 ASR affected concrete structures 222–5, 224f assessment: definition 6 ASTM standards 19, 46, 49–50, 55, 59, 87, 93, 98, 106, 124, 140, 141, 157, 163–4, 164t, 165, 167–8, 167f, 175, 200 Austin, S. et al. 167 autogenous shrinkage 154 Bartlett, F.M. 123 Bazant, Z.P. 19 Beattie, A.G. 99 Bentur, A. et al. 38, 39, 42 Berke, N.S. et al. 215 Bertolini, L. et al. 39, 40, 43, 47, 56, 58 Bilger, W. et al. 76–7, 78 bleeding 11–12, 11f bond strength: failure patterns 171; field testing methods 169–71, 169f;
laboratory testing methods 165–9, 166f bonding agents 187, 230 Bray, D.E. 86, 98–9, 107, 117, 118 bridge assessment 132–8; bridge classification 133–4; concrete delamination 140; environmental attack 134; inspection levels 137–8; materials deterioration 135; rating 136–7; structural damage 135–6 bridge deck repair 225–30; crack reinforcement 228; preventive maintenance 225–7; spall repair 228–30; see also GFRPs for bridge deck replacement British Standards 59, 123, 164, 197 Broomfield, J.P. et al. 47, 143–4 building and infrastructure degradation 2–5 building dynamics 116–18 Bungey, J.H. 92, 93, 111, 123, 144 Calleja, J. 52 Campbell-Allen, D. 36, 52, 78 carbon fiber reinforced polymers (CFRPs) 241 carbonation-induced corrosion 38–41, 212 Carino, N.J. 93, 97, 98, 106, 107 cathodic protection (CP) 215–19 cavitation 36–7 cementitious repair materials 171–4 Chandler, I. 4–5, 79 chemical stability 161 Chen, W.F. 65, 71, 72, 74, 133, 137 chloride diffusion test 32–5, 33f, 34f chloride-induced corrosion 41–3, 42t, 212
Index 343 chloride removal techniques 220, 221–2, 221f classical lamination theory 251, 255–66, 256f, 257f, 264f cohesive failure 171 compatibility of new and old materials 152–63; chemical stability 161; dimensional stability 153–60; transport properties 161–2 composite laminates: classical lamination theory 251, 255–66, 256f, 257f, 264f; elastic behavior 252–5, 253f, 254f; failure 269–73 compressive test on cored specimen 121–4, 122f concrete columns: strengthening 287–8, 287f, 321–7; behavior of FRP strengthened column/failure mode 321–2, 322f, 323f; confining effect 323–6, 324f, 325f; design of FRPstrengthened column 326, 327f concrete delamination 140 Concrete Society 123, 140 concrete strength assessment 121–7, 149; compressive test on cored specimen 121–4, 122f; rebound hammer measurement 124–6, 125f; surface wave measurement 127, 128f condition surveys 81–3 construction records 149 corrosion-damaged reinforced structures 178, 211–22; carbonationinduced 38–41, 212; cathodic protection methods 215–19; chlorideinduced 41–3, 42t, 212; chloride removal techniques 220, 221–2, 221f; realkalization techniques 219–20, 220f; restoration of surrounding concrete 7, 212–15, 213f, 214f crack repairs in concrete 196–204, 197t, 198t, 203f; evaluation 139, 139f; reinforcement 228; resin injection 199–203, 202f, 203f; stitching 203, 204, 204f creep 17–19 curing and protection 14, 14t Davidson, N. et al. 112 Dawson, J.L. et al. 143 degradation of reinforced concrete structures 2–5, 8–75, 177–9; chemical causes 27; disasters 62–75; lack of durability 25–61; mechanical causes 27; non-uniform dimensional changes
9–19; physical causes 27; repeated loading 19–25, 20f, 21f, 22f; visual symptoms 9 design of beams strengthened in flexure: design framework 304–14, 306f, 307f, 308f; general assumptions 301; initial situation of member 303–4, 304f; material behavior 301, 302–3, 303t destructive testing 84 dimensional stability 153–60; shrinkage 153–6, 155t; thermal expansion 9–11, 10f, 156–8, 157t, 158f; uneven stress distribution 158, 159–60, 159f Dowrick, D. 69, 71 drying shrinkage 14, 15–17, 154, 155–6, 155t Duan, L. 133, 137 durability problems 25–61; acid attack 55–6; alkali-silica reaction (ASR) 47–51; basic influencing factors 28–35, 29t; causes of deterioration 27; chloride diffusion test 32–5, 33f, 34f; definition 26; frost attack 56–9, 57f; marine environment 59–61, 60f; permeability tests 30–2; physical action 35–7; steel corrosion 37–47; sulfate attack 51–5, 53t earthquakes 65–75, 66f; bridge damage 135–6; building design 73–5, 74f, 75f; causes 65–9; intensity 69, 71–3; magnitude 69–71 electrical and electro-chemical testing 102–6, 103f, 103t, 104f, 105f, 107f electrical impedance 162 electromagnetic wave technique 109–16; infrared thermography 112–14, 114f; radar 111–12; radiometry and radiography 114–16 environmental and loading conditions 77–9, 78f epoxy-bonded steel plates 238–41 epoxy concrete 175 epoxy resins 199–200, 215 erosion 36 European Standards 164–5, 165t evaluation: definition 6 FHWA 136 fiber reinforced polymers (FRP) 241–2, 280–9; advantages and limitations 288–9; materials for fabrication 280–1, 281t; strengthening principles
344
Index
and configurations 282–5, 282f, 283f, 285f; strengthening process 285–8, 287f filament winding 248, 248f FIP 84, 147 fire damage 62–5, 204–9; assessment 129–32, 207; fire-damage factors 132; heat penetration 131–2; maximum temperature 130–1, 130t, 205–7, 205f, 206f, 207f; repair methods 209–11; strength loss 207–8, 208f, 209f; structural damage 205 flexural strengthening of beams 282, 282f 290–301; failure modes 290–2, 291f, 292f, 293f; load capacity 292, 293–9, 294f, 294t, 295f, 295t, 297f, 297t, 298t; maximum FRP strain 299–301, 300f, 301f; see also design of beams strengthened in flexure flexural tests 166f, 168–9 frost attack 56–9, 57f FRP see fiber reinforced polymers GFRPs (glass fiber-reinforced polymers) 241 GFRPs for bridge deck replacement 243–77; analysis of composite sections 273–7; (equivalent elastic properties 273–5; stress and deformation analysis 275–7, 275f); analysis of FRP deck members 251–73, 251f; (classical lamination theory 251, 255–66, 256f, 257f, 264f; elastic behavior of composite lamina 252–5, 253f, 254f; equivalent forces and moment 266–9; failure of composite laminates 269–73); connections 251; filament winding 248, 248f; hand lay-up 248, 249–50, 249f; materials 244–5, 244t; pultrusion 245–7, 246f, 247f; vacuum-assisted resin transfer molding (VARTM) 250, 250f Gouda, V.K. 42 grinding 191–2 Gutenberg, B. 70 gypsum-based concrete 173 half-cell measurement 103f, 140–2 Halicka, A. 165 Hall, J.F. 75 Halpin, J.C. 251 hand lay-up 248, 249–50, 249f Hassanen, M.A.H. 298
Hausman, D.A. 42 Hisano, M. et al. 75 Hu, Y-X. et al. 71, 75 hydration of concrete 10–11 hydrodemolition 229 ICC (International Code Council) 148 ICRI (International Concrete Repair Institute) 150–1, 152, 154 impact–echo method 95–8, 96f impressed current system 217–19, 217f, 219f infrared thermography 112–14, 114f infrastructure: definition 6–7 inspection and evaluation 76–146; bridge assessment 132–8; concrete strength assessment 121–7, 149; definition 7; detailed investigation 83–6; environmental and loading conditions 77–9, 78f; fire damage assessment 129–32, 207; non-destructive tests 85, 86–118, 149; preliminary investigations 79–83, 148–50; purpose 76–9; reflected and transmitted waves 118–21, 119f, 120f; reinforced steel corrosion assessment 138–46; structure classes 77, 78, 78f; surface cracking measurement 127, 128–9 interlaminar failure 272–3, 272f, 273f Jones, R.M. 251 Kay, T. 84, 90, 124, 126, 141, 151 Keller, T. 244 Kelly, J.M. 75 Khalifa, A. et al. 317 Klieger, P. et al. 126 Knab, L.I. 169, 170 Kolek, J. 126 Krol, M. 165 Kuneida, M. et al. 168 laminated composites see classical lamination theory latex modified cement systems 176–7 Leung, C.K.Y. et al. 299–300, 318 L’Hermite 14 Li, W. 104 Li, Y. 165 Li, Z. et al. 49, 73, 100, 101, 104, 113, 144 Limbachiya, M.C. 174 Lin, J.M. 98
Index 345 linear polarization resistance (LPR) 142–4, 143f, 143t linseed oil 226 lithium technologies 223–5, 224f loading 19–25, 20f, 21f, 22f, 77–9, 78f Long, A.E. 169 McBride, D. 117, 118 MacGregor, J.G. 123 McLeish, A. 170 magnesium phosphate concrete 173–4 magnetic flux leakage (MFL) 107, 108–9 magnetic particle testing 109 magnetic technique 106, 107–9, 108f Mailvaganam, N.P. 9, 35, 45, 55, 64, 81, 89, 97, 100, 126, 153, 162–3 Maji, A.K. 100 Malek, A.M. 305, 306 Malhotra, H.L. 63, 64 Malhotra, V.M. 93, 97, 106, 107 Mall, G. 169 Mangat, P.S. 174 Manitoba deck system 249–50, 249f marine environment 59–61, 60f Mathey, R.G. 170 maximum strain theory 270–1 maximum stress theory 269–70 mechanical wave techniques (MWT) 88–102; acoustic emission (AE) 98–102, 99f, 144–6, 145f; impact– echo method 95–8, 96f; ultrasonic testing 89–95, 91f, 92f, 94f Mehta, P.K. 18, 26, 49, 54, 58, 61 Meinheit, D.F. 176 methyl methacrylate concrete 175–6 Millard, S.G. 92, 93, 111, 123, 144 milling machines 229 Mindess, S. et al. 18 Modena, C. 318 Monon, J.F. 176 Monteiro, P.J.M. 18, 26, 49, 58 Monti, G. 326 Murray, A. 169 National Bridge Inspection Standard (NBIS) 133 National Research Council 25–6 NCHRP 350 Report 197 Neville, A.M. et al. 19, 32, 131 Newman, A. 5, 90, 94, 106, 109, 112, 198–9 non-destructive tests (NDT) 86–118, 149; building dynamics 116–18; definition 85; electrical and electro-
chemical methods 102–6, 103f, 103t, 104f, 105f, 107f; electromagnetic wave technique 109–16; magnetic technique 106, 107–9, 108f; mechanical wave techniques (MWT) 88–102; terminology 86; testing objects and problems 87–8 non-invasive testing 84–5 nuclear magnetic resonance (NMR) 109 Page, C.L. 40–1, 45 patching repair 192–6, 192f, 195f Patel, K. 305, 306 pavements 231–5, 231f; concrete overlay 232–3; full-depth repair 233–5 Pellegrino, C. 318 Perkins, P.H. 172, 239 permeability tests 30–2 Placido, F. 131 plastic shrinkage 12–14, 13f, 153–4 pneumatic breakers 229 polymer modified repair materials 174–7 polyurethane concrete 176 polyurethane resins 200–1 Popovics, S. 61 Portland cement-based concrete (PCC) 26, 27, 52, 53, 55, 171–3 Powers, T.C. et al. 29, 29t preliminary investigations 79–83, 148–50; condition survey 81–3, 149; desk study 79–81 preventive maintenance 225–7 pull-off tests 169–70, 169f Pullar-Strecker, P. 237 pultrusion 244–7, 246f, 247f radar 111–12 radiometry and radiography 114–16 Raoof, M. 298 realkalization 219–20, 220f rebound hammer measurement 124–6, 125f reflected and transmitted waves 118–21, 119f, 120f rehabilitation: definition 7 renovation engineering 1, 5–6 repair 7, 147, 148–52 repair materials 152–77; bond strength 165–71, 166f, 169f; cementitious materials 171–4; compatibility requirements 152–63; moduli of elasticity 160, 160t; polymer modified
346
Index
materials 174–7; standard testing methods 163–5 repair techniques 177–235; ASRinduced damage 222–5, 224f; bridge decks 225–30; corrosion damage 178, 211–22; cracks in concrete 196–204, 197t, 198t, 203f; fire-damaged concrete 204–11; general strategies 179–80; pavements 231–5, 231f; surface preparation 180–7, 182t; surface repair 188–96 resin injection 199–203, 202f, 203f restoration: definition 7 retrofitting: definition 7 Richter, C.F. 70 Rizzo, E.M. 169 Roper, H. 36, 52, 78 Ryan, K.L. 75 sacrificial anode system 215–16, 216f Sansalone, M. 97, 98 Scawthorn, C. 65, 71, 72, 74 scour 136 SCRG (steel cord reinforced grout) 241–2 SCRPs (steel cord reinforced polymers) 241–2 Sebastian, W.M. 299 serviceability: definition 26 Shah, S.P. 100, 101 shear strengthening of beams 282–3, 283f, 314–21; design of shearstrengthened beams 319–21, 320f; failure modes 314–17, 315f, 316f; maximum FRP strip spacing 321; shear contribution of FRP reinforcement 317–19 shear tests 166–7, 166f shotcrete 188–90, 189f shrinkage of concrete 12–17, 153–6, 155t silicone 226 slant-shear test 166f, 167–8, 167f Sobelman, M. 169 spall repair 228–30 Spoelstra, M.R. 326 sprayed concrete 188–90, 189f Stanley, R.K. 86, 98–9, 107 steel corrosion 37–47, 178, 178f; carbonation-induced corrosion 38–41; chloride-induced corrosion 41–3, 42t; corrosion mechanisms 43–5, 44t; protection strategies 45–7
steel corrosion assessment 138–46; acoustic emission (AE) method 144–6, 145f; half-cell measurement 103f, 140–2; linear polarization resistance (LPR) measurement 142–4, 143f, 143t; visual inspection and delamination survey 138–40, 139f steel, hot-rolled reinforcing 208, 210t Stehno, G. 169 stitching 203, 204, 204f Stokes, D.B. 223 strengthening 147–52, 241–2; see also concrete columns; design of beams strengthened in flexure; fiber reinforced polymers (FRP); flexural strengthening of beams; shear strengthening of beams strengthening techniques 235–42, 278–9; addition of reinforcing steel 236–7; design considerations 235–6; epoxy-bonded steel plates 238–41; externally bonded fiber reinforced sheets 241–2; replacement of reinforcement 237–8 stress distribution 158, 159–60, 159t structure classes 77, 78, 78f sulfate attack 51–5, 53t surface coatings 227 surface cracking measurement 127, 128–9 surface preparation 180–7, 182t; bonding agents 187, 230; chemical methods 184–5; flame cleaning 185; mechanical methods 183–4; surface requirement 185–6 surface repair 188–96; grinding 191–2; patching 192–6, 192f, 195f; replacement 191; shotcrete 188–90, 189f surface sealers 225–7 surface wave measurement 127, 128f Teng, J.G. et al. 293, 326 tension tests 166f, 168–9, 168f tensor polynomial failure criterion 271 thermal expansion 9–11, 10f, 156–8, 157t, 158t Trabanelli, G. 47 transport properties 161–2 Treadway, K.W.J. 45 Triantafillou, T.C. 317 Troxell, G.E. et al. 14 Tsai-Hill theory 271
Index 347 ultrasonic testing 89–95, 91f, 92f, 94f vacuum-assisted resin transfer molding (VARTM) 250, 250f Valenta, O. 31 Vassie, P.R. 106
Xanthakos, P.P. 137 Xi, Y. 165, 190 Yang, Q. et al. 174 Zhang, J. et al. 94